mirror of
https://github.com/mulgadc/spinifex.git
synced 2026-11-06 19:01:10 +00:00
Compare commits
1 Commits
| Author | SHA1 | Date | |
|---|---|---|---|
 Ben Duncan
|
caf6c438d2 |
@@ -1,3 +0,0 @@
|
||||
# The committed bundle carries the AWS SDK's template strings verbatim,
|
||||
# including trailing spaces that are part of their values.
|
||||
spinifex/runtime/roles/spinifexui/frontend/dist/** -whitespace
|
||||
@@ -0,0 +1,15 @@
|
||||
# CODEOWNERS — GitHub PR review enforcement
|
||||
#
|
||||
# Format: <path pattern> <owner> [<owner> …]
|
||||
# Order matters — last matching pattern wins.
|
||||
#
|
||||
# Set the matching branch protection rule on `dev` and `main` to
|
||||
# "Require review from Code Owners" or this file is advisory only.
|
||||
|
||||
# Network layer is governed by its invariant tests.
|
||||
# Changes here must respect those clauses
|
||||
# see spinifex/domains/network/invariants/
|
||||
# mulga docs/development/proposals/region-local-architecture/0006-spinifex-network-layer-contract.md (proposal, not accepted)
|
||||
/spinifex/domains/network/ @joshsiv-mulga
|
||||
/spinifex/runtime/roles/vpcd/ @joshsiv-mulga
|
||||
/spinifex/domains/network/invariants/ @joshsiv-mulga
|
||||
Binary file not shown.
|
Before Width: | Height: | Size: 167 KiB |
@@ -19,8 +19,8 @@
|
||||
</defs>
|
||||
<rect width="900" height="480" fill="url(#bg)" rx="12"/>
|
||||
|
||||
<text x="24" y="30" font-family="'Segoe UI', system-ui, sans-serif" font-size="11" fill="#a1a1aa" letter-spacing="2.5" font-weight="600">NGINX + ALB · PUBLIC SUBNETS · SPREAD ACROSS NODES</text>
|
||||
<text x="876" y="30" font-family="'SF Mono', ui-monospace, monospace" font-size="10" fill="#71717a" text-anchor="end">10.20.0.0/16 · server_count = 3</text>
|
||||
<text x="24" y="30" font-family="'Segoe UI', system-ui, sans-serif" font-size="11" fill="#a1a1aa" letter-spacing="2.5" font-weight="600">NGINX + ALB · VPC WITH PUBLIC AND PRIVATE SUBNETS</text>
|
||||
<text x="876" y="30" font-family="'SF Mono', ui-monospace, monospace" font-size="10" fill="#71717a" text-anchor="end">10.20.0.0/16 · multi-AZ</text>
|
||||
|
||||
<!-- Internet -->
|
||||
<rect x="350" y="54" width="200" height="40" rx="8" fill="url(#inet-grad)" stroke="#3fb950" stroke-width="1.1" stroke-opacity="0.55"/>
|
||||
@@ -41,45 +41,58 @@
|
||||
<line x1="230" y1="184" x2="230" y2="202" stroke="#DC2626" stroke-width="1" stroke-dasharray="3,3" opacity="0.6"/>
|
||||
<line x1="670" y1="184" x2="670" y2="202" stroke="#DC2626" stroke-width="1" stroke-dasharray="3,3" opacity="0.6"/>
|
||||
|
||||
<!-- public_a -->
|
||||
<g transform="translate(108, 202)">
|
||||
<rect width="244" height="150" rx="8" fill="url(#pub-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<rect width="244" height="104" rx="8" fill="url(#pub-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="20" y="24" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#3fb950" font-weight="700">public_a</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">10.20.1.0/24</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">az-a</text>
|
||||
<line x1="16" y1="32" x2="228" y2="32" stroke="#3fb950" stroke-opacity="0.3" stroke-width="0.5"/>
|
||||
<rect x="22" y="42" width="200" height="22" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="122" y="57" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">ALB ENI</text>
|
||||
<rect x="22" y="70" width="200" height="36" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="122" y="86" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">nginx-alb-1</text>
|
||||
<text x="122" y="100" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">public IP · Ubuntu · cloud-init</text>
|
||||
<rect x="22" y="110" width="200" height="36" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="122" y="126" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">nginx-alb-3</text>
|
||||
<text x="122" y="140" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">public IP · Ubuntu · cloud-init</text>
|
||||
|
||||
<rect x="20" y="42" width="200" height="22" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="120" y="57" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">ALB ENI</text>
|
||||
|
||||
<rect x="20" y="70" width="200" height="22" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="120" y="85" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">NAT GW</text>
|
||||
</g>
|
||||
|
||||
<!-- public_b -->
|
||||
<g transform="translate(548, 202)">
|
||||
<rect width="244" height="150" rx="8" fill="url(#pub-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<rect width="244" height="104" rx="8" fill="url(#pub-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="20" y="24" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#3fb950" font-weight="700">public_b</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">10.20.2.0/24</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">az-b</text>
|
||||
<line x1="16" y1="32" x2="228" y2="32" stroke="#3fb950" stroke-opacity="0.3" stroke-width="0.5"/>
|
||||
<rect x="22" y="42" width="200" height="22" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="122" y="57" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">ALB ENI</text>
|
||||
<rect x="22" y="70" width="200" height="36" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="122" y="86" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">nginx-alb-2</text>
|
||||
<text x="122" y="100" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">public IP · Ubuntu · cloud-init</text>
|
||||
|
||||
<rect x="20" y="56" width="200" height="22" rx="4" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="120" y="71" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle" font-weight="600">ALB ENI</text>
|
||||
</g>
|
||||
|
||||
<!-- spread placement group -->
|
||||
<rect x="108" y="372" width="684" height="88" rx="8" fill="url(#priv-grad)" stroke="#a855f7" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="128" y="394" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#c084fc" font-weight="700">nginx-alb-spread</text>
|
||||
<text x="772" y="394" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#d8b4fe" text-anchor="end">placement group · strategy spread · one server per node</text>
|
||||
<line x1="124" y1="402" x2="776" y2="402" stroke="#a855f7" stroke-opacity="0.3" stroke-width="0.5"/>
|
||||
<rect x="128" y="412" width="204" height="36" rx="5" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="230" y="428" font-family="'SF Mono', ui-monospace, monospace" font-size="11" fill="#d8b4fe" text-anchor="middle" font-weight="700">node1</text>
|
||||
<text x="230" y="442" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">one web server</text>
|
||||
<rect x="348" y="412" width="204" height="36" rx="5" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="450" y="428" font-family="'SF Mono', ui-monospace, monospace" font-size="11" fill="#d8b4fe" text-anchor="middle" font-weight="700">node2</text>
|
||||
<text x="450" y="442" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">one web server</text>
|
||||
<rect x="568" y="412" width="204" height="36" rx="5" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="670" y="428" font-family="'SF Mono', ui-monospace, monospace" font-size="11" fill="#d8b4fe" text-anchor="middle" font-weight="700">node3</text>
|
||||
<text x="670" y="442" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">one web server</text>
|
||||
<!-- SNAT arrow from public_a NAT → private row -->
|
||||
<line x1="230" y1="306" x2="230" y2="340" stroke="#F87171" stroke-width="1.2" stroke-dasharray="3,3" opacity="0.65"/>
|
||||
<polygon points="226,336 234,336 230,344" fill="#F87171" opacity="0.8"/>
|
||||
<rect x="256" y="314" width="240" height="20" rx="10" fill="#0a0a0a" stroke="#F87171" stroke-width="0.6" stroke-opacity="0.55"/>
|
||||
<text x="376" y="328" font-family="'SF Mono', ui-monospace, monospace" font-size="10" fill="#FCA5A5" text-anchor="middle">SNAT for private subnets (egress)</text>
|
||||
|
||||
<!-- private_a -->
|
||||
<g transform="translate(108, 350)">
|
||||
<rect width="244" height="104" rx="8" fill="url(#priv-grad)" stroke="#a855f7" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="20" y="24" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#c084fc" font-weight="700">private_a</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#d8b4fe" text-anchor="end">az-a</text>
|
||||
<line x1="16" y1="32" x2="228" y2="32" stroke="#a855f7" stroke-opacity="0.3" stroke-width="0.5"/>
|
||||
|
||||
<rect x="20" y="44" width="200" height="48" rx="5" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="120" y="64" font-family="'SF Mono', ui-monospace, monospace" font-size="11" fill="#d8b4fe" text-anchor="middle" font-weight="700">nginx_1</text>
|
||||
<text x="120" y="80" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">Debian 12 · cloud-init</text>
|
||||
</g>
|
||||
|
||||
<!-- private_b -->
|
||||
<g transform="translate(548, 350)">
|
||||
<rect width="244" height="104" rx="8" fill="url(#priv-grad)" stroke="#a855f7" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="20" y="24" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#c084fc" font-weight="700">private_b</text>
|
||||
<text x="224" y="24" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#d8b4fe" text-anchor="end">az-b</text>
|
||||
<line x1="16" y1="32" x2="228" y2="32" stroke="#a855f7" stroke-opacity="0.3" stroke-width="0.5"/>
|
||||
|
||||
<rect x="20" y="44" width="200" height="48" rx="5" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="120" y="64" font-family="'SF Mono', ui-monospace, monospace" font-size="11" fill="#d8b4fe" text-anchor="middle" font-weight="700">nginx_2</text>
|
||||
<text x="120" y="80" font-family="'Segoe UI', sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">Debian 12 · cloud-init</text>
|
||||
</g>
|
||||
</svg>
|
||||
|
||||
|
Before Width: | Height: | Size: 7.5 KiB After Width: | Height: | Size: 7.1 KiB |
@@ -1,110 +0,0 @@
|
||||
<svg xmlns="http://www.w3.org/2000/svg" width="900" height="580" viewBox="0 0 900 580">
|
||||
<defs>
|
||||
<linearGradient id="bg" x1="0%" y1="0%" x2="100%" y2="100%">
|
||||
<stop offset="0%" style="stop-color:#0a0a0a"/>
|
||||
<stop offset="100%" style="stop-color:#0d1117"/>
|
||||
</linearGradient>
|
||||
<linearGradient id="vpc-grad" x1="0%" y1="0%" x2="0%" y2="100%">
|
||||
<stop offset="0%" style="stop-color:#58a6ff14"/>
|
||||
<stop offset="100%" style="stop-color:#58a6ff05"/>
|
||||
</linearGradient>
|
||||
<linearGradient id="node-grad" x1="0%" y1="0%" x2="0%" y2="100%">
|
||||
<stop offset="0%" style="stop-color:#3fb9501f"/>
|
||||
<stop offset="100%" style="stop-color:#3fb95008"/>
|
||||
</linearGradient>
|
||||
<linearGradient id="svc-grad" x1="0%" y1="0%" x2="0%" y2="100%">
|
||||
<stop offset="0%" style="stop-color:#a855f71f"/>
|
||||
<stop offset="100%" style="stop-color:#a855f708"/>
|
||||
</linearGradient>
|
||||
<marker id="arrow" viewBox="0 0 10 10" refX="8" refY="5" markerWidth="7" markerHeight="7" orient="auto-start-reverse">
|
||||
<path d="M0,0 L10,5 L0,10 z" fill="#71717a"/>
|
||||
</marker>
|
||||
</defs>
|
||||
<rect width="900" height="580" fill="url(#bg)" rx="12"/>
|
||||
<text x="24" y="30" font-family="'Segoe UI', system-ui, sans-serif" font-size="11" fill="#a1a1aa" letter-spacing="2.5" font-weight="600">SPINIFEX ON VULTR · ONE NODE OR THREE</text>
|
||||
<text x="876" y="30" font-family="'SF Mono', ui-monospace, monospace" font-size="10" fill="#71717a" text-anchor="end">--nodes 1 or 3 · vx1-g-16c-64g or bare metal</text>
|
||||
<rect x="30" y="52" width="250" height="56" rx="8" fill="#e4e4e70d" stroke="#e4e4e7" stroke-width="1.1" stroke-opacity="0.6"/>
|
||||
<text x="155.0" y="76" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#e4e4e7" text-anchor="middle" font-weight="700">Your machine</text>
|
||||
<text x="155.0" y="94" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#a1a1aa" text-anchor="middle">deploy-vultr.sh · OpenTofu or Terraform</text>
|
||||
<rect x="325" y="52" width="250" height="56" rx="8" fill="#3fb9500d" stroke="#3fb950" stroke-width="1.1" stroke-opacity="0.6"/>
|
||||
<text x="450.0" y="76" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#3fb950" text-anchor="middle" font-weight="700">Internet</text>
|
||||
<text x="450.0" y="94" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#7ee787" text-anchor="middle">customers reach your EC2 and load balancers</text>
|
||||
<rect x="620" y="52" width="250" height="56" rx="8" fill="#58a6ff12" stroke="#58a6ff" stroke-width="1.1" stroke-opacity="0.6"/>
|
||||
<text x="745.0" y="76" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#79c0ff" text-anchor="middle" font-weight="700">Vultr API</text>
|
||||
<text x="745.0" y="94" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#a5d6ff" text-anchor="middle">VPC · servers · reserved IPs</text>
|
||||
<rect x="30" y="160" width="840" height="400" rx="10" fill="url(#vpc-grad)" stroke="#58a6ff" stroke-width="1.2" stroke-opacity="0.55" stroke-dasharray="6,4"/>
|
||||
<text x="50" y="184" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#79c0ff" font-weight="700">Vultr VPC</text>
|
||||
<text x="850" y="184" font-family="'SF Mono', ui-monospace, monospace" font-size="10" fill="#58a6ff" text-anchor="end">private network 10.10.8.0/24</text>
|
||||
<line x1="155" y1="108" x2="155" y2="158" stroke="#71717a" stroke-width="1.2" opacity="0.8" marker-end="url(#arrow)"/>
|
||||
<text x="165" y="138" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#a1a1aa" text-anchor="start">Terraform builds · SSH installs</text>
|
||||
<line x1="450" y1="108" x2="450" y2="158" stroke="#71717a" stroke-width="1.2" opacity="0.8" marker-end="url(#arrow)"/>
|
||||
<text x="460" y="138" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#a1a1aa" text-anchor="start">reserved IPs</text>
|
||||
<line x1="745" y1="158" x2="745" y2="110" stroke="#71717a" stroke-width="1.2" opacity="0.8" marker-end="url(#arrow)"/>
|
||||
<text x="735" y="138" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#a1a1aa" text-anchor="end">Spinifex calls the API</text>
|
||||
<rect x="60" y="205" width="250" height="118" rx="8" fill="url(#node-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.6"/>
|
||||
<text x="78" y="229" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#3fb950" font-weight="700">node1</text>
|
||||
<text x="292" y="229" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">Ubuntu 24.04 · Spinifex</text>
|
||||
<line x1="74" y1="237" x2="296" y2="237" stroke="#3fb950" stroke-opacity="0.3" stroke-width="0.6"/>
|
||||
<rect x="76" y="250" width="74" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="113.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">AWS API</text>
|
||||
<rect x="160" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="190.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">EBS</text>
|
||||
<rect x="230" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="260.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">S3</text>
|
||||
<text x="185" y="300" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">VPC networking · your instances</text>
|
||||
<rect x="325" y="205" width="250" height="118" rx="8" fill="url(#node-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.6" stroke-dasharray="5,3"/>
|
||||
<text x="343" y="229" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#3fb950" font-weight="700">node2</text>
|
||||
<text x="557" y="229" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">with --nodes 3</text>
|
||||
<line x1="339" y1="237" x2="561" y2="237" stroke="#3fb950" stroke-opacity="0.3" stroke-width="0.6"/>
|
||||
<rect x="341" y="250" width="74" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="378.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">AWS API</text>
|
||||
<rect x="425" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="455.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">EBS</text>
|
||||
<rect x="495" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="525.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">S3</text>
|
||||
<text x="450" y="300" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">VPC networking · your instances</text>
|
||||
<rect x="590" y="205" width="250" height="118" rx="8" fill="url(#node-grad)" stroke="#3fb950" stroke-width="1.2" stroke-opacity="0.6" stroke-dasharray="5,3"/>
|
||||
<text x="608" y="229" font-family="'SF Mono', ui-monospace, monospace" font-size="12.5" fill="#3fb950" font-weight="700">node3</text>
|
||||
<text x="822" y="229" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#7ee787" text-anchor="end">with --nodes 3</text>
|
||||
<line x1="604" y1="237" x2="826" y2="237" stroke="#3fb950" stroke-opacity="0.3" stroke-width="0.6"/>
|
||||
<rect x="606" y="250" width="74" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="643.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">AWS API</text>
|
||||
<rect x="690" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="720.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">EBS</text>
|
||||
<rect x="760" y="250" width="60" height="24" rx="5" fill="#3fb9501a" stroke="#3fb950" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="790.0" y="266" font-family="'SF Mono', ui-monospace, monospace" font-size="10.5" fill="#7ee787" text-anchor="middle">S3</text>
|
||||
<text x="715" y="300" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#a1a1aa" text-anchor="middle">VPC networking · your instances</text>
|
||||
<line x1="310" y1="264" x2="325" y2="264" stroke="#58a6ff" stroke-width="1.6" stroke-dasharray="3,3" opacity="0.9"/>
|
||||
<line x1="575" y1="264" x2="590" y2="264" stroke="#58a6ff" stroke-width="1.6" stroke-dasharray="3,3" opacity="0.9"/>
|
||||
<text x="450" y="348" font-family="'Segoe UI', system-ui, sans-serif" font-size="10" fill="#79c0ff" text-anchor="middle">nodes cluster over the private network: control plane, EBS replicas and S3 shards</text>
|
||||
<rect x="60" y="370" width="780" height="166" rx="8" fill="url(#svc-grad)" stroke="#a855f7" stroke-width="1.2" stroke-opacity="0.55"/>
|
||||
<text x="80" y="394" font-family="'SF Mono', ui-monospace, monospace" font-size="12" fill="#c084fc" font-weight="700">AWS services across the fleet</text>
|
||||
<text x="820" y="394" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#d8b4fe" text-anchor="end">spread across the nodes; with three, moved to a survivor if one fails</text>
|
||||
<line x1="76" y1="402" x2="824" y2="402" stroke="#a855f7" stroke-opacity="0.3" stroke-width="0.6"/>
|
||||
<rect x="76" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="119.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">EC2</text>
|
||||
<text x="119.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">instances</text>
|
||||
<rect x="169" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="212.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">ALB / NLB</text>
|
||||
<text x="212.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">load balancers</text>
|
||||
<rect x="262" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="305.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">EKS</text>
|
||||
<text x="305.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">Kubernetes</text>
|
||||
<rect x="355" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="398.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">ECS</text>
|
||||
<text x="398.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">containers</text>
|
||||
<rect x="448" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="491.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">RDS</text>
|
||||
<text x="491.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">databases</text>
|
||||
<rect x="541" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="584.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">S3</text>
|
||||
<text x="584.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">object storage</text>
|
||||
<rect x="634" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="677.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">EBS</text>
|
||||
<text x="677.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">volumes</text>
|
||||
<rect x="727" y="418" width="86" height="52" rx="6" fill="#a855f71a" stroke="#a855f7" stroke-opacity="0.5" stroke-width="0.6"/>
|
||||
<text x="770.0" y="440" font-family="'SF Mono', ui-monospace, monospace" font-size="11.5" fill="#d8b4fe" text-anchor="middle" font-weight="700">IAM</text>
|
||||
<text x="770.0" y="457" font-family="'Segoe UI', system-ui, sans-serif" font-size="9" fill="#a1a1aa" text-anchor="middle">access</text>
|
||||
<text x="450" y="500" font-family="'Segoe UI', system-ui, sans-serif" font-size="10.5" fill="#a1a1aa" text-anchor="middle">Use the AWS CLI, SDKs and Terraform against your own endpoint, unmodified.</text>
|
||||
<text x="450" y="520" font-family="'Segoe UI', system-ui, sans-serif" font-size="9.5" fill="#71717a" text-anchor="middle">Public addresses are Vultr reserved IPs, attached to the node each resource runs on.</text>
|
||||
</svg>
|
||||
|
Before Width: | Height: | Size: 13 KiB |
Binary file not shown.
|
Before Width: | Height: | Size: 725 KiB |
Binary file not shown.
|
Before Width: | Height: | Size: 743 KiB |
Binary file not shown.
|
After Width: | Height: | Size: 151 KiB |
+5
-31
@@ -8,8 +8,7 @@
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-single",
|
||||
"extra": "",
|
||||
"suite": "",
|
||||
"timeout_minutes": 60
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 2,
|
||||
@@ -20,8 +19,7 @@
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-single",
|
||||
"extra": "",
|
||||
"suite": "",
|
||||
"timeout_minutes": 60
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 3,
|
||||
@@ -32,8 +30,7 @@
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-single",
|
||||
"extra": "",
|
||||
"suite": "",
|
||||
"timeout_minutes": 60
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 8,
|
||||
@@ -168,7 +165,7 @@
|
||||
"runner": "ci-multi",
|
||||
"extra": "",
|
||||
"suite": "instancerecovery",
|
||||
"timeout_minutes": 125
|
||||
"timeout_minutes": 105
|
||||
},
|
||||
{
|
||||
"cell": 32,
|
||||
@@ -203,8 +200,7 @@
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-single",
|
||||
"extra": "",
|
||||
"suite": "",
|
||||
"timeout_minutes": 60
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 35,
|
||||
@@ -227,27 +223,5 @@
|
||||
"runner": "ci-single",
|
||||
"extra": "nat-single",
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 37,
|
||||
"install": "binary",
|
||||
"topo": "real-multi",
|
||||
"net": "dhcp",
|
||||
"host": "oracle-10",
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-multi",
|
||||
"extra": "",
|
||||
"suite": ""
|
||||
},
|
||||
{
|
||||
"cell": 38,
|
||||
"install": "binary",
|
||||
"topo": "real-multi",
|
||||
"net": "nat",
|
||||
"host": "oracle-10",
|
||||
"arch": "x86_64",
|
||||
"runner": "ci-multi",
|
||||
"extra": "nat-multi",
|
||||
"suite": ""
|
||||
}
|
||||
]
|
||||
|
||||
@@ -6,6 +6,16 @@
|
||||
|
||||
<!-- The notable changes, as bullets. -->
|
||||
|
||||
## Account-scoped lifecycle impact
|
||||
|
||||
<!-- Complete this section only when the PR changes account-scoped resource lifecycle, account termination, resource cleanup, or a lifecycle boundary. Otherwise write: N/A. -->
|
||||
|
||||
- Lifecycle owner and resource family:
|
||||
- Owner-local ordering, recovery, retained-state, or force impact:
|
||||
- Cross-domain dependency or ownership impact:
|
||||
- Evidence added or rechecked:
|
||||
- ADR-0008 lifecycle register / ADR-0001 migration-record / bead updates:
|
||||
|
||||
## Community Contribution License
|
||||
|
||||
- [ ] All community contributions in this pull request are licensed to Mulga under the terms of the [Apache License 2.0](https://www.apache.org/licenses/LICENSE-2.0). By creating this pull request I represent that I have the right to license these contributions to Mulga under the Apache License 2.0.
|
||||
|
||||
@@ -1,148 +0,0 @@
|
||||
#!/usr/bin/env bash
|
||||
# e2e-select.sh push <before> <sha> | suites "<e2e-x ...>"
|
||||
# Prints the e2e gate's single_legs, multi_suites and multi_timeout outputs.
|
||||
# Exits non-zero on anything it cannot read, which fails the e2e status.
|
||||
set -euo pipefail
|
||||
|
||||
SCRIPTS_DIR=$(CDPATH='' cd -- "$(dirname -- "$0")" && pwd)
|
||||
# shellcheck source=suite-sets.sh
|
||||
. "${SCRIPTS_DIR}/suite-sets.sh"
|
||||
|
||||
# Provision, install and teardown around the suites; the floor is what the
|
||||
# jobs have always had, so a budget can only be raised from it.
|
||||
SETUP_ALLOWANCE_MIN=15
|
||||
FLOOR_MIN=60
|
||||
|
||||
BASE_RUNNER=ci-single
|
||||
SERVICE_RUNNER=ci-pseudo
|
||||
|
||||
log() { printf '[e2e-select] %s\n' "$*" >&2; }
|
||||
die() { log "error: $*"; exit 1; }
|
||||
|
||||
# in_list <word> <list>
|
||||
in_list() {
|
||||
local w
|
||||
for w in $2; do [ "$w" = "$1" ] && return 0; done
|
||||
return 1
|
||||
}
|
||||
|
||||
# keep <list> <eligible>: the words of <list> that are in <eligible>, deduplicated.
|
||||
keep() {
|
||||
local out="" s
|
||||
for s in $1; do
|
||||
in_list "$s" "$2" || continue
|
||||
in_list "$s" "$out" || out="${out:+$out }$s"
|
||||
done
|
||||
printf '%s' "$out"
|
||||
}
|
||||
|
||||
# budget <list>: max over the suites' own timeouts plus the setup allowance.
|
||||
# A sum would demand hours for a set that shares one job and never takes them.
|
||||
budget() {
|
||||
local want="${FLOOR_MIN}" s t
|
||||
for s in $1; do
|
||||
t="$(e2e_suite_timeout "$s")"
|
||||
t=$((${t%m} + SETUP_ALLOWANCE_MIN))
|
||||
[ "$t" -gt "$want" ] && want="$t"
|
||||
done
|
||||
printf '%s' "$want"
|
||||
}
|
||||
|
||||
prefixed() {
|
||||
local out="" s
|
||||
for s in $1; do out="${out:+$out }e2e-$s"; done
|
||||
printf '%s' "$out"
|
||||
}
|
||||
|
||||
# leg <name> <runner> <suites>: one matrix entry, or nothing for no suites.
|
||||
leg() {
|
||||
[ -n "$3" ] || return 0
|
||||
jq -nc --arg leg "$1" --arg runner "$2" --arg suites "$(prefixed "$3")" \
|
||||
--argjson timeout "$(budget "$3")" \
|
||||
'{leg: $leg, runner: $runner, suites: $suites, timeout: $timeout}'
|
||||
}
|
||||
|
||||
# emit <base-single> <service-single> <multi>
|
||||
emit() {
|
||||
local legs
|
||||
legs=$( { leg base "${BASE_RUNNER}" "$1"; leg service "${SERVICE_RUNNER}" "$2"; } | jq -sc .)
|
||||
log "base leg: ${1:-none}"
|
||||
log "service leg: ${2:-none}"
|
||||
log "multi: ${3:-none}"
|
||||
printf 'single_legs=%s\n' "$legs"
|
||||
printf 'multi_suites=%s\n' "$(prefixed "$3")"
|
||||
printf 'multi_timeout=%s\n' "$(budget "$3")"
|
||||
}
|
||||
|
||||
# select_for_paths <newline-separated paths> [all]: base set plus matches.
|
||||
# With "all", every mapped suite is selected, for a push with no baseline.
|
||||
select_for_paths() {
|
||||
local matched="" p s
|
||||
if [ "${2:-}" = all ]; then
|
||||
matched=$(e2e_path_suites_all | tr '\n' ' ')
|
||||
else
|
||||
while IFS= read -r p; do
|
||||
[ -n "$p" ] || continue
|
||||
for s in $(e2e_suites_for_path "$p"); do
|
||||
in_list "$s" "$matched" || { matched="${matched:+$matched }$s"; log "${p} -> ${s}"; }
|
||||
done
|
||||
done <<<"$1"
|
||||
fi
|
||||
# A matched suite runs where it is eligible, preferring one node, so route53
|
||||
# goes to the service leg and lb to the multi job.
|
||||
local extra_single="" extra_multi=""
|
||||
for s in $matched; do
|
||||
if in_list "$s" "${E2E_SUITES_SINGLE}"; then
|
||||
extra_single="${extra_single:+$extra_single }$s"
|
||||
elif in_list "$s" "${E2E_SUITES_MULTI}"; then
|
||||
extra_multi="${extra_multi:+$extra_multi }$s"
|
||||
else
|
||||
die "E2E_PATH_SUITES names '${s}', which no topology can run"
|
||||
fi
|
||||
done
|
||||
local base_single base_multi
|
||||
base_single=$(keep "${E2E_SUITES_PUSH_BASE}" "${E2E_SUITES_SINGLE}")
|
||||
base_multi=$(keep "${E2E_SUITES_PUSH_BASE}" "${E2E_SUITES_MULTI}")
|
||||
# A matched suite already in the base leg runs there, not twice.
|
||||
extra_single=$(for s in $extra_single; do in_list "$s" "$base_single" || printf '%s ' "$s"; done)
|
||||
emit "$base_single" "$(keep "$extra_single" "${E2E_SUITES_SINGLE}")" \
|
||||
"$(keep "$base_multi $extra_multi" "${E2E_SUITES_MULTI}")"
|
||||
}
|
||||
|
||||
mode="${1:-}"
|
||||
case "$mode" in
|
||||
push)
|
||||
[ $# -eq 3 ] || die "usage: $0 push <before> <sha>"
|
||||
before="$2" sha="$3"
|
||||
git cat-file -e "${sha}^{commit}" 2>/dev/null || die "commit ${sha} is not in this checkout"
|
||||
if [ -z "${before//0/}" ]; then
|
||||
log "no previous commit for this push; selecting every mapped suite"
|
||||
select_for_paths "" all
|
||||
exit 0
|
||||
fi
|
||||
if ! git cat-file -e "${before}^{commit}" 2>/dev/null; then
|
||||
git fetch -q --filter=blob:none origin "${before}" 2>/dev/null || true
|
||||
fi
|
||||
if ! git cat-file -e "${before}^{commit}" 2>/dev/null; then
|
||||
log "previous commit ${before} is gone; selecting every mapped suite"
|
||||
select_for_paths "" all
|
||||
exit 0
|
||||
fi
|
||||
# A rename counts as both paths, so moving code out of a service selects it.
|
||||
changed=$(git diff --no-renames --name-only "${before}" "${sha}") || die "git diff ${before} ${sha} failed"
|
||||
log "$(printf '%s\n' "$changed" | grep -c . || true) paths changed ${before:0:9}..${sha:0:9}"
|
||||
select_for_paths "$changed"
|
||||
;;
|
||||
suites)
|
||||
[ $# -eq 2 ] || die "usage: $0 suites \"<e2e-x ...>\""
|
||||
if [ -z "${2// /}" ]; then
|
||||
select_for_paths ""
|
||||
exit 0
|
||||
fi
|
||||
requested=$(printf '%s' "$2" | sed 's/e2e-//g')
|
||||
emit "$(keep "$requested" "${E2E_SUITES_SINGLE}")" "" "$(keep "$requested" "${E2E_SUITES_MULTI}")"
|
||||
;;
|
||||
*)
|
||||
die "usage: $0 push <before> <sha> | suites \"<list>\""
|
||||
;;
|
||||
esac
|
||||
@@ -1,45 +0,0 @@
|
||||
#!/usr/bin/env bash
|
||||
# Trims GOCACHE to the entries one CI run used. Go refreshes an entry's mtime
|
||||
# on use only when it is over an hour old, so "mark" backdates every restored
|
||||
# entry first and "trim" drops whatever this run neither read nor wrote.
|
||||
set -euo pipefail
|
||||
|
||||
cache="$(go env GOCACHE)"
|
||||
mark="${RUNNER_TEMP:?}/go-build-cache.mark"
|
||||
|
||||
# stale prints each entry this run did not touch and no touched action entry
|
||||
# names. Go skips rewriting an output it already holds, so a fresh action
|
||||
# entry can point at a backdated output.
|
||||
stale() {
|
||||
local keep="${RUNNER_TEMP}/go-build-cache.keep"
|
||||
find "$cache" -mindepth 2 -maxdepth 2 -name '*-a' -newer "$mark" -exec cat {} + |
|
||||
awk '{ print "/" $3 "-d" }' >"$keep"
|
||||
find "$cache" -mindepth 2 -maxdepth 2 ! -newer "$mark" | { grep -vFf "$keep" || true; }
|
||||
}
|
||||
|
||||
case "${1:-}" in
|
||||
mark)
|
||||
mkdir -p "$cache"
|
||||
find "$cache" -mindepth 2 -maxdepth 2 -exec touch -m -d '2 hours ago' {} +
|
||||
touch "$mark"
|
||||
;;
|
||||
stale | trim)
|
||||
if [[ ! -f "$mark" ]]; then
|
||||
echo "go-build-cache: no mark, leaving the cache untrimmed" >&2
|
||||
exit 0
|
||||
fi
|
||||
if [[ "$1" == stale ]]; then
|
||||
stale
|
||||
exit 0
|
||||
fi
|
||||
before=$(du -sm "$cache" | cut -f1)
|
||||
# An entry is a file, or a directory holding one cached executable whose
|
||||
# mtime Go refreshes in place of the binary's, so it goes as a unit.
|
||||
stale | xargs -r rm -rf --
|
||||
echo "trimmed ${before}MB -> $(du -sm "$cache" | cut -f1)MB"
|
||||
;;
|
||||
*)
|
||||
echo "usage: $0 mark|stale|trim" >&2
|
||||
exit 2
|
||||
;;
|
||||
esac
|
||||
@@ -1,66 +0,0 @@
|
||||
#!/usr/bin/env bash
|
||||
# Usage: hypervisor-disk-preflight.sh <env.tfvars>
|
||||
#
|
||||
# Fails fast when a hypervisor named in the tofu env is short of disk, before
|
||||
# `tofu apply` tries to define a domain there. libvirt writes each domain's XML
|
||||
# under /etc/libvirt/qemu on the root filesystem and its disks into the pool, and
|
||||
# a full one fails the apply minutes in as "Domain Creation Failed".
|
||||
#
|
||||
# Hosts are reached the way the libvirt provider reaches them (ssh <user>@<name>,
|
||||
# resolved through ~/.ssh/config). A host that cannot be reached is reported and
|
||||
# skipped: tofu's own error for it is clearer than anything said here.
|
||||
#
|
||||
# Env: HV_SSH_USER (tf-user), HV_MIN_ROOT_FREE_MB (1024), HV_MIN_POOL_FREE_GB (20).
|
||||
set -uo pipefail
|
||||
|
||||
tfvars="${1:?usage: $0 <env.tfvars>}"
|
||||
user="${HV_SSH_USER:-tf-user}"
|
||||
min_root_kb=$(( ${HV_MIN_ROOT_FREE_MB:-1024} * 1024 ))
|
||||
min_pool_kb=$(( ${HV_MIN_POOL_FREE_GB:-20} * 1024 * 1024 ))
|
||||
|
||||
if [ ! -r "$tfvars" ]; then
|
||||
echo "::error::hypervisor disk preflight: cannot read $tfvars" >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
# One node per line in every env file: { name = "...", ..., storage_pool = "..." }.
|
||||
pairs=$(grep -v '^[[:space:]]*#' "$tfvars" |
|
||||
sed -nE 's/.*name[[:space:]]*=[[:space:]]*"([^"]+)".*storage_pool[[:space:]]*=[[:space:]]*"([^"]+)".*/\1 \2/p')
|
||||
if [ -z "$pairs" ]; then
|
||||
echo "::warning::hypervisor disk preflight: no nodes with a storage_pool in $tfvars; nothing checked"
|
||||
exit 0
|
||||
fi
|
||||
|
||||
# Prints "<root-avail-kb> <pool-avail-kb> <pool-path>" for the pool named in $1.
|
||||
# shellcheck disable=SC2016 # expanded on the hypervisor, not here
|
||||
remote='set -u
|
||||
root=$(df -Pk /etc/libvirt/qemu | awk "NR==2{print \$4}")
|
||||
path=$(virsh -c qemu:///system pool-dumpxml "$1" 2>/dev/null | sed -n "s:.*<path>\(.*\)</path>.*:\1:p" | head -n1)
|
||||
pool=-
|
||||
[ -n "$path" ] && pool=$(df -Pk "$path" | awk "NR==2{print \$4}")
|
||||
echo "$root $pool ${path:--}"'
|
||||
|
||||
fail=0
|
||||
while read -r host pool; do
|
||||
[ -n "$host" ] || continue
|
||||
if ! out=$(ssh -o BatchMode=yes -o ConnectTimeout=10 "${user}@${host}" bash -s -- "$pool" <<<"$remote" 2>&1); then
|
||||
echo "::warning::hypervisor disk preflight: cannot reach $host ($out); its disk is unchecked"
|
||||
continue
|
||||
fi
|
||||
read -r root_kb pool_kb pool_path <<<"$(tail -n1 <<<"$out")"
|
||||
if ! [[ "$root_kb" =~ ^[0-9]+$ ]]; then
|
||||
echo "::warning::hypervisor disk preflight: unexpected answer from $host: $out"
|
||||
continue
|
||||
fi
|
||||
echo "$host: root fs $((root_kb / 1024)) MiB free; pool $pool ($pool_path) $([[ "$pool_kb" =~ ^[0-9]+$ ]] && echo "$((pool_kb / 1024 / 1024)) GiB free" || echo "not found")"
|
||||
if [ "$root_kb" -lt "$min_root_kb" ]; then
|
||||
echo "::error::hypervisor $host has $((root_kb / 1024)) MiB free on the filesystem holding /etc/libvirt/qemu (need ${HV_MIN_ROOT_FREE_MB:-1024} MiB): libvirt cannot write a domain definition there. Free space on $host before re-running." >&2
|
||||
fail=1
|
||||
fi
|
||||
if [[ "$pool_kb" =~ ^[0-9]+$ ]] && [ "$pool_kb" -lt "$min_pool_kb" ]; then
|
||||
echo "::error::hypervisor $host has $((pool_kb / 1024 / 1024)) GiB free in storage pool $pool at $pool_path (need ${HV_MIN_POOL_FREE_GB:-20} GiB): the node's disk cannot grow. Free space on $host before re-running." >&2
|
||||
fail=1
|
||||
fi
|
||||
done <<<"$pairs"
|
||||
|
||||
exit "$fail"
|
||||
@@ -66,67 +66,6 @@ E2E_SUITES_MULTI="multinode lb cert quota route53 kvquorum lbrecovery instancere
|
||||
E2E_SUITES_SINGLE_RE="$(printf '%s' "${E2E_SUITES_SINGLE}" | tr ' ' '|')"
|
||||
E2E_SUITES_MULTI_RE="$(printf '%s' "${E2E_SUITES_MULTI}" | tr ' ' '|')"
|
||||
|
||||
# The push gate always runs this set. A push adds to it the suites whose code
|
||||
# it changed, by E2E_PATH_SUITES below; a dispatch names its own set instead.
|
||||
E2E_SUITES_PUSH_BASE="single multinode cert iam partialblock"
|
||||
|
||||
# "suite prefix" pairs: a push changing a path under prefix adds suite. The
|
||||
# recovery and fault suites stay nightly, and shared code adds nothing because
|
||||
# the base set covers it. Image directories carry each guest's own scripts.
|
||||
E2E_PATH_SUITES="
|
||||
rds cmd/rds-agent/
|
||||
rds internal/rdsgw/
|
||||
rds spinifex/agents/rds/
|
||||
rds spinifex/domains/rds/
|
||||
rds tests/e2e/rds/
|
||||
rds scripts/images/rds-mariadb/
|
||||
rds scripts/images/rds-postgres/
|
||||
ecs cmd/ecs-agent/
|
||||
ecs contracts/ecs/
|
||||
ecs internal/ecsgw/
|
||||
ecs spinifex/agents/ecs/
|
||||
ecs spinifex/domains/ecs/
|
||||
ecs tests/e2e/ecs/
|
||||
ecs scripts/images/ecs-agent/
|
||||
eks cmd/eks-
|
||||
eks contracts/eks/
|
||||
eks internal/eksgw/
|
||||
eks spinifex/agents/eks/
|
||||
eks spinifex/domains/eks/
|
||||
eks tests/e2e/eks/
|
||||
eks scripts/images/eks-node/
|
||||
rds scripts/images/common/
|
||||
ecs scripts/images/common/
|
||||
eks scripts/images/common/
|
||||
rds scripts/build-system-image.sh
|
||||
ecs scripts/build-system-image.sh
|
||||
eks scripts/build-system-image.sh
|
||||
route53 spinifex/domains/route53/
|
||||
route53 tests/e2e/route53/
|
||||
lb cmd/lb-agent/
|
||||
lb spinifex/agents/elbv2/
|
||||
lb spinifex/domains/elbv2/
|
||||
lb tests/e2e/lb/
|
||||
"
|
||||
|
||||
# Every suite E2E_PATH_SUITES can name, once each, in table order.
|
||||
# e2e_path_suites_all
|
||||
e2e_path_suites_all() {
|
||||
printf '%s\n' "${E2E_PATH_SUITES}" | awk 'NF == 2 && !seen[$1]++ { print $1 }'
|
||||
}
|
||||
|
||||
# The suites a change to <path> triggers, one per line.
|
||||
# e2e_suites_for_path <path>
|
||||
e2e_suites_for_path() {
|
||||
local path="$1" suite prefix
|
||||
while read -r suite prefix; do
|
||||
[ -n "${prefix}" ] || continue
|
||||
case "${path}" in
|
||||
"${prefix}"*) printf '%s\n' "${suite}" ;;
|
||||
esac
|
||||
done <<<"${E2E_PATH_SUITES}"
|
||||
}
|
||||
|
||||
# What every e2e-nightly permutation cell must prove, deliberately narrower
|
||||
# than the two sets above: those cells have a 35-minute budget and exist to show
|
||||
# that every install / network / host-OS combination boots and serves. The
|
||||
@@ -178,9 +117,8 @@ e2e_suite_timeout() {
|
||||
# storage case waits for a pause measured at 4m47s and then holds four
|
||||
# minutes to prove nothing moved. Three tests, each taking a node away and
|
||||
# bringing it back, and the partition case adds a lease validity, a lease
|
||||
# TTL and a boot on top of everything the other two wait for. The data
|
||||
# test lets its guest age past the pending watchdog twice, about 20 more.
|
||||
instancerecovery) echo "110m" ;;
|
||||
# TTL and a boot on top of everything the other two wait for.
|
||||
instancerecovery) echo "90m" ;;
|
||||
# lbrecovery pays for a load balancer twice: once to stand one up with two
|
||||
# backends placed around it, and once for the recovery itself — a stale
|
||||
# owner, a lease, a boot, HAProxy's own start, and its health checks against
|
||||
@@ -191,10 +129,6 @@ e2e_suite_timeout() {
|
||||
# budget on a node stopping and starting: seven units down, a raft election,
|
||||
# and the whole service stack back up with every stream caught up.
|
||||
kvquorum) echo "40m" ;;
|
||||
# single is the broadest suite and finishes, it does not hang: passing runs
|
||||
# took 24-29m, and an Oracle host ran out at 30m inside the last test's
|
||||
# cleanup. The default left it no headroom at all.
|
||||
single) echo "40m" ;;
|
||||
*) echo "30m" ;;
|
||||
esac
|
||||
}
|
||||
|
||||
@@ -1,34 +0,0 @@
|
||||
#!/usr/bin/env bash
|
||||
# Usage: wait-for-memory.sh <MiB>
|
||||
#
|
||||
# Waits until the runner's MemAvailable reaches <MiB>, then returns. Runners on
|
||||
# one host share its RAM, and a neighbour's build holds several GiB for a few
|
||||
# minutes, so one sample taken during it is not a verdict on the host.
|
||||
# Gives up after MEM_WAIT_TIMEOUT_S and names the biggest processes.
|
||||
#
|
||||
# Env: MEM_WAIT_TIMEOUT_S (900), MEM_WAIT_POLL_S (30), MEMINFO (/proc/meminfo).
|
||||
set -uo pipefail
|
||||
|
||||
want_mb="${1:?usage: $0 <MiB>}"
|
||||
timeout_s="${MEM_WAIT_TIMEOUT_S:-900}"
|
||||
poll_s="${MEM_WAIT_POLL_S:-30}"
|
||||
meminfo="${MEMINFO:-/proc/meminfo}"
|
||||
|
||||
avail_mb() { awk '/^MemAvailable:/ {print int($2 / 1024)}' "$meminfo"; }
|
||||
|
||||
start=$SECONDS
|
||||
while :; do
|
||||
have=$(avail_mb)
|
||||
if [ -n "$have" ] && [ "$have" -ge "$want_mb" ]; then
|
||||
echo "memory: ${have} MiB available, need ${want_mb} MiB"
|
||||
exit 0
|
||||
fi
|
||||
waited=$(( SECONDS - start ))
|
||||
if [ "$waited" -ge "$timeout_s" ]; then
|
||||
echo "::error::need ${want_mb} MiB available, have ${have:-unknown} MiB after waiting ${waited}s"
|
||||
ps -eo rss,pid,args --sort=-rss | head -11 | cut -c1-200
|
||||
exit 1
|
||||
fi
|
||||
echo "memory: ${have:-unknown} MiB available, need ${want_mb} MiB; waiting (${waited}s of ${timeout_s}s)"
|
||||
sleep "$poll_s"
|
||||
done
|
||||
@@ -18,11 +18,6 @@ workflows:
|
||||
- 'actions/create-github-app-token@v3.3.0'
|
||||
- 'actions/upload-artifact@v7.0.2'
|
||||
- 'hashicorp/setup-terraform@v4.0.1'
|
||||
'.github/workflows/e2e-cloudvendor-vultr.yml':
|
||||
- 'actions/checkout@v7.0.1'
|
||||
- 'actions/create-github-app-token@v3.3.0'
|
||||
- 'actions/upload-artifact@v7.0.2'
|
||||
- 'hashicorp/setup-terraform@v4.0.1'
|
||||
'.github/workflows/e2e-ddil.yml':
|
||||
- 'actions/checkout@v7.0.1'
|
||||
- 'actions/create-github-app-token@v3.3.0'
|
||||
@@ -78,7 +73,6 @@ workflows:
|
||||
'.github/workflows/ui.yml':
|
||||
- 'actions/checkout@v7.0.1'
|
||||
- 'actions/setup-node@v7.1.0'
|
||||
- 'actions/upload-artifact@v7.0.2'
|
||||
- 'pnpm/action-setup@v6.1.0'
|
||||
dependencies:
|
||||
'actions/cache@v6.1.0':
|
||||
|
||||
@@ -1,21 +0,0 @@
|
||||
# Codex review gate: calls bluebottle's reusable workflows from main, so a PR cannot change its own review.
|
||||
name: codex-review
|
||||
|
||||
on:
|
||||
pull_request_target:
|
||||
branches: [dev]
|
||||
types: [opened, synchronize, reopened, ready_for_review, edited]
|
||||
issue_comment:
|
||||
types: [created]
|
||||
|
||||
permissions: {}
|
||||
|
||||
jobs:
|
||||
review:
|
||||
if: github.event_name == 'pull_request_target'
|
||||
uses: mulgadc/bluebottle/.github/workflows/codex-pr-review.yml@main
|
||||
secrets: inherit
|
||||
waive:
|
||||
if: github.event_name == 'issue_comment' && github.event.issue.pull_request
|
||||
uses: mulgadc/bluebottle/.github/workflows/codex-waive.yml@main
|
||||
secrets: inherit
|
||||
@@ -30,10 +30,6 @@ concurrency:
|
||||
jobs:
|
||||
prerelease:
|
||||
runs-on: ubuntu-26.04
|
||||
env:
|
||||
# Dev prereleases, newest published first: GitHub lists releases by tag name
|
||||
# as text, so list order is not build order. The install Worker uses this rule.
|
||||
DEV_RELEASES_JQ: '[.[] | select(.draft == false and .prerelease and .published_at != null and (.tag_name | test("-dev(\\.|$)")))] | sort_by(.published_at) | reverse'
|
||||
steps:
|
||||
# Full history and tags, because the tag name is git describe and a shallow
|
||||
# clone has no tag to describe against.
|
||||
@@ -106,9 +102,9 @@ jobs:
|
||||
env:
|
||||
GH_TOKEN: ${{ github.token }}
|
||||
|
||||
- name: Build microVM artifacts (lb-agent + kernel + initramfs + qboot + SQS store and worker images)
|
||||
- name: Build microVM artifacts (lb-agent + kernel + initramfs + qboot)
|
||||
if: steps.version.outputs.exists != 'true'
|
||||
run: cd spinifex && make build-microvm-image build-sqs-worker-microvm-image
|
||||
run: cd spinifex && make build-microvm-image
|
||||
|
||||
- name: Build amd64 Tarball
|
||||
if: steps.version.outputs.exists != 'true'
|
||||
@@ -168,12 +164,12 @@ jobs:
|
||||
exit 1
|
||||
fi
|
||||
echo "latest is still $latest"
|
||||
dev="$(gh api "repos/$GITHUB_REPOSITORY/releases?per_page=100" --jq "$DEV_RELEASES_JQ | .[0].tag_name")"
|
||||
dev="$(gh api "repos/$GITHUB_REPOSITORY/releases?per_page=20" --jq 'map(select(.prerelease))[0].tag_name')"
|
||||
[ "$dev" = "$TAG" ] || { echo "::error::the dev channel resolves to $dev, not $TAG"; exit 1; }
|
||||
|
||||
# Kept rather than unbounded: the Worker reads one page of releases to resolve
|
||||
# the channel, so enough prereleases would eventually push the newest off it.
|
||||
# Five leaves a few recent builds to roll back to.
|
||||
# Kept rather than unbounded: the Worker reads the first page of releases to
|
||||
# resolve the channel, so enough prereleases would eventually push the real
|
||||
# ones off it. Five leaves a few recent builds to roll back to.
|
||||
- name: Prune old dev prereleases
|
||||
if: steps.version.outputs.exists != 'true'
|
||||
env:
|
||||
@@ -181,7 +177,7 @@ jobs:
|
||||
run: |
|
||||
set -euo pipefail
|
||||
gh api "repos/$GITHUB_REPOSITORY/releases?per_page=100" \
|
||||
--jq "$DEV_RELEASES_JQ | .[5:] | .[].tag_name" \
|
||||
--jq 'map(select(.prerelease and (.tag_name | endswith("-dev")))) | .[5:] | .[].tag_name' \
|
||||
> stale.txt
|
||||
while read -r tag; do
|
||||
[ -n "$tag" ] || continue
|
||||
|
||||
@@ -71,16 +71,9 @@ jobs:
|
||||
ARTIFACT_DIR: /tmp/spinifex-e2e-nightly-cell-${{ inputs.cell || '20' }}-${{ github.run_id }}
|
||||
SPINIFEX_OTEL_ATTRS: ${{ inputs.otel-attrs || format('ci.run_id={0},ci.workflow={1},ci.cell={2},ci.matrix=baremetal/pxe/real-hw', github.run_id, github.workflow, inputs.cell || '20') }}
|
||||
steps:
|
||||
# A cancelled job's root steps outlive it, since the runner cannot reach
|
||||
# them once it kills sudo, and would keep writing here and to the cache.
|
||||
# A rootfs killed before bin/bash exists is one the build refuses to reuse.
|
||||
- name: Clean Workspace
|
||||
run: |
|
||||
sudo systemctl stop spinifex-bm-pxe.scope spinifex-bm-e2e.scope 2>/dev/null || true
|
||||
sudo umount -R -l "${BM_ISO_CACHE}/rootfs" 2>/dev/null || true
|
||||
if [ -d "${BM_ISO_CACHE}/rootfs" ] && [ ! -x "${BM_ISO_CACHE}/rootfs/bin/bash" ]; then
|
||||
sudo rm -rf --one-file-system "${BM_ISO_CACHE}/rootfs"
|
||||
fi
|
||||
if [ ! -c /dev/null ]; then sudo rm -f /dev/null; sudo mknod -m 666 /dev/null c 1 3; fi
|
||||
sudo rm -rf --one-file-system "$GITHUB_WORKSPACE"/*
|
||||
|
||||
@@ -120,15 +113,10 @@ jobs:
|
||||
"$GITHUB_WORKSPACE" \
|
||||
"${RUNNER_TEMP}/spinifex-tests.tar" \
|
||||
"${RUNNER_TEMP}/bm-bins"
|
||||
# Built here, as the runner user, so the PXE build finds it current.
|
||||
make -C "$GITHUB_WORKSPACE" build-microvm-image
|
||||
|
||||
- name: Build PXE Artifacts
|
||||
run: |
|
||||
# -o build-spinifex: the binaries come from Build Artifacts, and a
|
||||
# rebuild here would write root-owned files into the workspace.
|
||||
sudo systemd-run --scope --collect --quiet --unit=spinifex-bm-pxe -- \
|
||||
make -C mulga/scripts/iso-builder -o build-spinifex \
|
||||
sudo make -C mulga/scripts/iso-builder \
|
||||
build-rootfs build-squashfs build-initramfs \
|
||||
CACHE="${BM_ISO_CACHE}" \
|
||||
SPINIFEX_BIN="${RUNNER_TEMP}/bm-bins/spx" \
|
||||
@@ -136,10 +124,6 @@ jobs:
|
||||
SPINIFEX_SRC="${GITHUB_WORKSPACE}" \
|
||||
VIPERBLOCK_SRC="${RUNNER_TEMP}/bm-bins/viperblock" \
|
||||
VERSION="$(cd "$GITHUB_WORKSPACE" && git describe --tags --always --exclude '*-dev' 2>/dev/null || echo dev)"
|
||||
if root_owned="$(find "$GITHUB_WORKSPACE" -user root -print -quit)" && [ -n "$root_owned" ]; then
|
||||
echo "::error::the PXE build wrote root-owned files into the workspace, e.g. $root_owned"
|
||||
exit 1
|
||||
fi
|
||||
|
||||
- name: Write ePDU credentials
|
||||
run: |
|
||||
@@ -211,18 +195,13 @@ jobs:
|
||||
done
|
||||
export BM_TEST_TIMEOUT
|
||||
echo "suites=${BM_SUITES} test-timeout=${BM_TEST_TIMEOUT}"
|
||||
sudo -E systemd-run --scope --collect --quiet --unit=spinifex-bm-e2e -- \
|
||||
mulga/scripts/baremetal/run-baremetal-e2e.sh \
|
||||
sudo -E mulga/scripts/baremetal/run-baremetal-e2e.sh \
|
||||
--vmlinuz "${BM_ISO_CACHE}/vmlinuz" \
|
||||
--initrd "${BM_ISO_CACHE}/initrd.img" \
|
||||
--squashfs "${BM_ISO_CACHE}/spinifex.squashfs" \
|
||||
--tests-tar "${RUNNER_TEMP}/spinifex-tests.tar" \
|
||||
2>&1 | tee "${ARTIFACT_DIR}/test-baremetal.log"
|
||||
|
||||
- name: Stop this job's root steps
|
||||
if: always()
|
||||
run: sudo systemctl stop spinifex-bm-pxe.scope spinifex-bm-e2e.scope 2>/dev/null || true
|
||||
|
||||
- name: Generate JUnit report
|
||||
if: always()
|
||||
run: |
|
||||
|
||||
@@ -1,142 +0,0 @@
|
||||
name: E2E Cloud Vendor Vultr
|
||||
|
||||
# On demand only. Deploys a ref to Vultr with deploy-vultr.sh, runs the same
|
||||
# published Terraform workbooks as the hypervisor nightly and OCI, and destroys
|
||||
# the deployment, failing if anything is left billing.
|
||||
on:
|
||||
workflow_dispatch:
|
||||
inputs:
|
||||
nodes:
|
||||
description: "1 or 3 nodes"
|
||||
required: false
|
||||
default: "1"
|
||||
type: choice
|
||||
options: ["1", "3"]
|
||||
plan:
|
||||
description: "Vultr plan: vx1-g-16c-64g (VM) or vbm-8c-64gb-amd (bare metal)"
|
||||
required: false
|
||||
default: vx1-g-16c-64g
|
||||
region:
|
||||
description: "Vultr region"
|
||||
required: false
|
||||
default: ord
|
||||
workbooks:
|
||||
description: "Space-separated workbooks (blank = the driver's full list)"
|
||||
required: false
|
||||
source:
|
||||
description: "tree (build this ref) or release (the published installer)"
|
||||
required: false
|
||||
default: tree
|
||||
type: choice
|
||||
options: [tree, release]
|
||||
channel:
|
||||
description: "With source=release: latest, or dev for the newest prerelease"
|
||||
required: false
|
||||
default: latest
|
||||
keep_on_fail:
|
||||
description: "Leave a failed deployment up to inspect. It keeps billing."
|
||||
required: false
|
||||
type: boolean
|
||||
default: false
|
||||
runner:
|
||||
description: "Runner label (blank = the VULTR_E2E_RUNNER variable, else ubuntu-26.04)"
|
||||
required: false
|
||||
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
# Deployments are named per run, but the account's reserved-IP quota is shared,
|
||||
# so a second dispatch queues rather than competing with the first.
|
||||
concurrency:
|
||||
group: e2e-vultr
|
||||
cancel-in-progress: false
|
||||
|
||||
jobs:
|
||||
vultr:
|
||||
# Hosted by default: the job only drives Terraform and SSH. The API key's
|
||||
# access control must admit this runner, so a hosted runner needs a key
|
||||
# whose list holds 0.0.0.0/0; a self-hosted runner can use a fixed address.
|
||||
runs-on: ${{ inputs.runner || vars.VULTR_E2E_RUNNER || 'ubuntu-26.04' }}
|
||||
timeout-minutes: 240
|
||||
env:
|
||||
VULTR_NAME: e2e-${{ github.run_id }}
|
||||
VULTR_NODES: ${{ inputs.nodes }}
|
||||
VULTR_PLAN: ${{ inputs.plan }}
|
||||
VULTR_REGION: ${{ inputs.region }}
|
||||
VULTR_SOURCE: ${{ inputs.source }}
|
||||
VULTR_CHANNEL: ${{ inputs.channel }}
|
||||
VULTR_ARTIFACT_DIR: ${{ github.workspace }}/e2e-vultr
|
||||
SPX_REF: ${{ github.ref_name }}
|
||||
steps:
|
||||
- uses: actions/checkout@v7.0.1
|
||||
with:
|
||||
path: spinifex
|
||||
fetch-depth: 0
|
||||
|
||||
- uses: actions/create-github-app-token@v3.3.0
|
||||
id: app-token
|
||||
with:
|
||||
client-id: ${{ secrets.APP_ID }}
|
||||
private-key: ${{ secrets.APP_PRIVATE_KEY }}
|
||||
repositories: mulga
|
||||
|
||||
# build-install-artifacts.sh lives in mulga; the workbooks and their driver
|
||||
# are in spinifex, so only the one directory is needed here.
|
||||
- uses: actions/checkout@v7.0.1
|
||||
with:
|
||||
repository: mulgadc/mulga
|
||||
path: mulga
|
||||
token: ${{ steps.app-token.outputs.token }}
|
||||
sparse-checkout: scripts/tofu-cluster
|
||||
|
||||
- uses: hashicorp/setup-terraform@v4.0.1
|
||||
with:
|
||||
terraform_wrapper: false
|
||||
|
||||
- name: Run Vultr
|
||||
working-directory: spinifex/scripts/terraform/vultr-spx
|
||||
env:
|
||||
VULTR_API_KEY: ${{ secrets.VULTR_API_KEY }}
|
||||
MULGA_ROOT: ${{ github.workspace }}/mulga
|
||||
# Through the environment, never interpolated into the script: a
|
||||
# dispatch input is untrusted text to the shell.
|
||||
WORKBOOKS_INPUT: ${{ inputs.workbooks }}
|
||||
VULTR_KEEP_ON_FAIL: ${{ inputs.keep_on_fail && '1' || '0' }}
|
||||
run: |
|
||||
set -uo pipefail
|
||||
if [ -n "$WORKBOOKS_INPUT" ]; then export WORKBOOKS="$WORKBOOKS_INPUT"; fi
|
||||
./e2e-vultr.sh 2>&1 | tee "$RUNNER_TEMP/e2e-vultr.log"
|
||||
exit "${PIPESTATUS[0]}"
|
||||
|
||||
# always(): a cancelled or timed-out job is the one that would leave
|
||||
# servers and reserved IPs billing. destroy works from this run's state.
|
||||
- name: Sweep anything left behind
|
||||
if: always() && !inputs.keep_on_fail
|
||||
working-directory: spinifex/scripts/terraform/vultr-spx
|
||||
env:
|
||||
VULTR_API_KEY: ${{ secrets.VULTR_API_KEY }}
|
||||
run: ./e2e-vultr.sh --sweep-only
|
||||
|
||||
- name: Publish the verdict
|
||||
if: always()
|
||||
run: |
|
||||
{
|
||||
echo "### Vultr: ${VULTR_NODES} x ${VULTR_PLAN} in ${VULTR_REGION}, ${VULTR_SOURCE} build of ${SPX_REF}"
|
||||
echo '```'
|
||||
cat "$VULTR_ARTIFACT_DIR/verdict.txt" 2>/dev/null || echo "no verdict written"
|
||||
echo '```'
|
||||
} >> "$GITHUB_STEP_SUMMARY"
|
||||
|
||||
# Logs and the verdict only. The throwaway SSH key and the Terraform state
|
||||
# under ~/.spinifex never reach an artifact.
|
||||
- name: Upload the run
|
||||
if: always()
|
||||
uses: actions/upload-artifact@v7.0.2
|
||||
with:
|
||||
name: e2e-vultr-${{ github.run_id }}
|
||||
path: |
|
||||
${{ runner.temp }}/e2e-vultr.log
|
||||
${{ github.workspace }}/e2e-vultr/*.log
|
||||
${{ github.workspace }}/e2e-vultr/verdict.txt
|
||||
${{ github.workspace }}/e2e-vultr/node*/*.log
|
||||
retention-days: 14
|
||||
@@ -6,32 +6,11 @@ name: E2E Nightly
|
||||
# `main` and `dev` as separate nightlies, and it would always execute `main`'s
|
||||
# copy of this file — leaving a fix to the workflow itself untestable until it
|
||||
# had already shipped. A dispatch takes both the file and the source from its ref.
|
||||
#
|
||||
# Cells (matrix cells are defined in .github/e2e-cells.json):
|
||||
#
|
||||
# Single node Three nodes
|
||||
# 1 Debian 13, DHCP 8 Debian 13, DHCP
|
||||
# 2 Ubuntu 26.04, static 9 Ubuntu 26.04, static
|
||||
# 3 Ubuntu 26.04, DHCP 10 Ubuntu 26.04, DHCP
|
||||
# 17 tofu examples 29 storage fault injection
|
||||
# 18 reboot 30 routed NAT
|
||||
# 19 routed NAT 31 instance recovery
|
||||
# 25 eks 32 load balancer recovery
|
||||
# 26 ecs 33 KV bucket quorum under node loss
|
||||
# 27 rds 37 Oracle Linux 10, DHCP
|
||||
# 34 Oracle Linux 10, DHCP 38 Oracle Linux 10, routed NAT
|
||||
# 35 Oracle Linux 10, reboot
|
||||
# 36 Oracle Linux 10, routed NAT
|
||||
#
|
||||
# Standalone jobs
|
||||
# 20 bare metal (PXE) 23 predastore throughput
|
||||
# 21 ISO, single node 24 predastore fault injection
|
||||
# 22 ISO, three nodes 28 guest disk performance (bare metal)
|
||||
on:
|
||||
workflow_dispatch:
|
||||
inputs:
|
||||
cells:
|
||||
description: "Comma-separated cell numbers, no spaces, e.g. 1,25,27 (blank = all). The cell list is at the top of .github/workflows/e2e-nightly.yml."
|
||||
description: "Comma-separated cell numbers to run, no spaces (blank = all). 1-3, 8-10, 17-19 tofu permutation matrix, 34-36 Oracle Linux single node (basic, reboot, routed NAT), 20 baremetal, 21-22 ISO, 23 predastore throughput, 24 predastore fault injection, 25-27 service suites (eks, ecs, rds), 28 guest disk performance, 29 storage fault injection, 30 routed NAT on three nodes, 31 instance recovery on three nodes, 32 load balancer recovery on three nodes, 33 KV bucket quorum under node loss on three nodes"
|
||||
required: false
|
||||
|
||||
permissions:
|
||||
@@ -205,14 +184,6 @@ jobs:
|
||||
echo "running qemu: $(pgrep -c qemu-system-x86 || echo 0)"
|
||||
} >> "${ARTIFACT_DIR}/hypervisor-load.txt"
|
||||
|
||||
# A hypervisor out of disk fails the apply minutes in, as "Domain Creation
|
||||
# Failed" from libvirt writing the domain XML. Say so before starting.
|
||||
- name: Preflight hypervisor disk
|
||||
working-directory: mulga/scripts/tofu-cluster
|
||||
run: |
|
||||
"${GITHUB_WORKSPACE}/spinifex/.github/scripts/hypervisor-disk-preflight.sh" \
|
||||
"envs/${SPINIFEX_CI_ENV}.tfvars"
|
||||
|
||||
- name: Provision VMs
|
||||
working-directory: mulga/scripts/tofu-cluster
|
||||
env:
|
||||
@@ -270,11 +241,6 @@ jobs:
|
||||
. "${GITHUB_WORKSPACE}/spinifex/.github/scripts/suite-sets.sh"
|
||||
e2e_suites_assert_subset "e2e-nightly single" "${E2E_SUITES_NIGHTLY_SINGLE}" "${E2E_SUITES_SINGLE}"
|
||||
|
||||
# The select job sized this cell's budget from these same timeouts, so
|
||||
# resolve them here and carry the pairs into the remote loop.
|
||||
SUITE_TMOS=""
|
||||
for s in ${E2E_SUITES_NIGHTLY_SINGLE}; do SUITE_TMOS="${SUITE_TMOS}${s}:$(e2e_suite_timeout "${s}") "; done
|
||||
|
||||
INVENTORY=$(tofu output -json inventory)
|
||||
WAN_IP=$(echo "$INVENTORY" | jq -r '.nodes[0].wan_ip')
|
||||
SSH_KEY=$(echo "$INVENTORY" | jq -r '.ssh_key_path')
|
||||
@@ -284,10 +250,9 @@ jobs:
|
||||
for SUITE in ${E2E_SUITES_NIGHTLY_SINGLE}; do \
|
||||
echo \"::group::suite \$SUITE\"; \
|
||||
date -u +%Y-%m-%dT%H:%M:%SZ > '${ARTIFACT_DIR}/test-'\$SUITE'.start'; \
|
||||
TMO=30m; for kv in ${SUITE_TMOS}; do case \$kv in \$SUITE:*) TMO=\${kv#*:} ;; esac; done; \
|
||||
GITHUB_ACTIONS=true FORCE_COLOR=1 SPINIFEX_E2E=1 SPINIFEX_MODE=single \
|
||||
ARTIFACT_DIR='${ARTIFACT_DIR}' \
|
||||
./\$SUITE.test -test.v -test.timeout=\$TMO \
|
||||
./\$SUITE.test -test.v -test.timeout=30m \
|
||||
2>&1 | tee '${ARTIFACT_DIR}/test-'\$SUITE'.log'; \
|
||||
rc=\${PIPESTATUS[0]}; echo \"::endgroup::\"; \
|
||||
if [ \$rc -ne 0 ]; then echo \"\$SUITE: FAILED\" >> '${ARTIFACT_DIR}/suite-status.txt'; FAIL=1; fi; \
|
||||
@@ -359,20 +324,16 @@ jobs:
|
||||
NODES_CSV="${NODES_CSV:+$NODES_CSV,}$ip"
|
||||
done
|
||||
|
||||
SUITE_TMOS=""
|
||||
for s in ${E2E_SUITES_NIGHTLY_MULTI}; do SUITE_TMOS="${SUITE_TMOS}${s}:$(e2e_suite_timeout "${s}") "; done
|
||||
|
||||
ssh -tt -o StrictHostKeyChecking=no -i "$SSH_KEY" tf-user@${NODE1_IP} \
|
||||
"set -u; mkdir -p '${ARTIFACT_DIR}'; cd \$HOME/e2e-tests/_bin; FAIL=0; \
|
||||
for SUITE in ${E2E_SUITES_NIGHTLY_MULTI}; do \
|
||||
echo \"::group::suite \$SUITE\"; \
|
||||
date -u +%Y-%m-%dT%H:%M:%SZ > '${ARTIFACT_DIR}/test-'\$SUITE'.start'; \
|
||||
TMO=30m; for kv in ${SUITE_TMOS}; do case \$kv in \$SUITE:*) TMO=\${kv#*:} ;; esac; done; \
|
||||
GITHUB_ACTIONS=true FORCE_COLOR=1 SPINIFEX_E2E=1 SPINIFEX_MODE=multinode \
|
||||
SPINIFEX_NODES='${NODES_CSV}' SPINIFEX_NODE_IPS='${NODES_CSV}' \
|
||||
SPINIFEX_SSH_USER=tf-user SPINIFEX_SSH_KEY=\$HOME/.ssh/tf-user-ap-southeast-2 \
|
||||
ARTIFACT_DIR='${ARTIFACT_DIR}' \
|
||||
./\$SUITE.test -test.v -test.timeout=\$TMO \
|
||||
./\$SUITE.test -test.v -test.timeout=30m \
|
||||
2>&1 | tee '${ARTIFACT_DIR}/test-'\$SUITE'.log'; \
|
||||
rc=\${PIPESTATUS[0]}; echo \"::endgroup::\"; \
|
||||
if [ \$rc -ne 0 ]; then echo \"\$SUITE: FAILED\" >> '${ARTIFACT_DIR}/suite-status.txt'; FAIL=1; fi; \
|
||||
@@ -1038,7 +999,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ env.GIT_BRANCH }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: mulga/spinifex
|
||||
|
||||
- uses: actions/checkout@v7.0.1
|
||||
@@ -1078,6 +1039,9 @@ jobs:
|
||||
{ echo "::error::br-wan not permitted in /etc/qemu/bridge.conf"; exit 1; }
|
||||
command -v sshpass >/dev/null ||
|
||||
{ echo "::error::sshpass not installed"; exit 1; }
|
||||
free_mb=$(free -m | awk '/^Mem:/ {print $7}')
|
||||
[ "$free_mb" -ge 8192 ] ||
|
||||
{ echo "::error::need 8 GiB available, have ${free_mb} MiB"; exit 1; }
|
||||
# A 40G image on tmpfs competes with the 8 GiB the VM itself needs.
|
||||
fstype=$(df --output=fstype "$TMPDIR" | tail -1 | tr -d ' ')
|
||||
[ "$fstype" != tmpfs ] ||
|
||||
@@ -1108,10 +1072,6 @@ jobs:
|
||||
ssh-keygen -t ed25519 -N "" -f "$HOME/.ssh/$k" -C "$k-ci"
|
||||
done
|
||||
|
||||
# The node's 8 GiB, claimed when the VM boots rather than at preflight.
|
||||
- name: Wait for memory
|
||||
run: mulga/spinifex/.github/scripts/wait-for-memory.sh 8192
|
||||
|
||||
# Boots the ISO, installs unattended, reads the lease off the serial
|
||||
# console, runs the in-guest smoke test, and installs the key for the
|
||||
# next step.
|
||||
@@ -1225,16 +1185,6 @@ jobs:
|
||||
sleep 2
|
||||
rm -f "$TMPDIR"/tmp.*.qcow2 "$TMPDIR"/tmp.*.log "$TMPDIR"/tmp.*-node.log
|
||||
|
||||
# The ISO build runs as root inside the workspace, and the next job on
|
||||
# this runner, from any workflow, cannot remove what it leaves there.
|
||||
- name: Release the workspace
|
||||
if: always()
|
||||
run: |
|
||||
sudo umount -R -l "$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/.cache/rootfs 2>/dev/null || true
|
||||
sudo rm -rf --one-file-system "$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/.cache \
|
||||
"$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/dist
|
||||
sudo chown -R "$(id -u):$(id -g)" "$GITHUB_WORKSPACE"
|
||||
|
||||
iso-multi:
|
||||
name: cell-22 (iso/dhcp/three-node)
|
||||
# After cell-21 rather than beside it: same concurrency group, and running
|
||||
@@ -1296,7 +1246,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ env.GIT_BRANCH }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: mulga/spinifex
|
||||
|
||||
- uses: actions/checkout@v7.0.1
|
||||
@@ -1323,10 +1273,15 @@ jobs:
|
||||
ref: ${{ steps.siblings.outputs.bluebottle }}
|
||||
path: mulga/bluebottle
|
||||
|
||||
# 3 x 8 GiB plus 9 sparse 40G disks, before the instances the cluster
|
||||
# launches on top.
|
||||
- name: Preflight
|
||||
run: |
|
||||
set -euo pipefail
|
||||
mkdir -p "$TMPDIR"
|
||||
free_mb=$(free -m | awk '/^Mem:/ {print $7}')
|
||||
[ "$free_mb" -ge 26624 ] ||
|
||||
{ echo "::error::need 26 GiB available, have ${free_mb} MiB"; exit 1; }
|
||||
fstype=$(df --output=fstype "$TMPDIR" | tail -1 | tr -d ' ')
|
||||
[ "$fstype" != tmpfs ] ||
|
||||
{ echo "::error::$TMPDIR is tmpfs — nine node disks would be in RAM"; exit 1; }
|
||||
@@ -1349,11 +1304,6 @@ jobs:
|
||||
# formation port. Both land with mulga-yxtda and the bridged WAN mode;
|
||||
# until then this cell proves the cluster forms and survives, not that a
|
||||
# customer can use it.
|
||||
# 3 x 8 GiB plus headroom for the instances the cluster launches on top,
|
||||
# claimed when the nodes boot rather than at preflight.
|
||||
- name: Wait for memory
|
||||
run: mulga/spinifex/.github/scripts/wait-for-memory.sh 26624
|
||||
|
||||
- name: Install, form and verify the cluster
|
||||
working-directory: mulga/scripts/iso-builder
|
||||
run: |
|
||||
@@ -1386,16 +1336,6 @@ jobs:
|
||||
rm -rf /tmp/cell22-artifacts-$GITHUB_RUN_ID /tmp/cell22-$GITHUB_RUN_ID.log \
|
||||
"$WORK_DIR"
|
||||
|
||||
# The ISO build runs as root inside the workspace, and the next job on
|
||||
# this runner, from any workflow, cannot remove what it leaves there.
|
||||
- name: Release the workspace
|
||||
if: always()
|
||||
run: |
|
||||
sudo umount -R -l "$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/.cache/rootfs 2>/dev/null || true
|
||||
sudo rm -rf --one-file-system "$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/.cache \
|
||||
"$GITHUB_WORKSPACE"/mulga/scripts/iso-builder/dist
|
||||
sudo chown -R "$(id -u):$(id -g)" "$GITHUB_WORKSPACE"
|
||||
|
||||
# Predastore throughput ("cell 23"). Predastore has no runners of its own and
|
||||
# spinifex depends on it, so the benchmark lives here and checks predastore
|
||||
# out as a sibling.
|
||||
@@ -1444,7 +1384,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Predastore
|
||||
@@ -1568,7 +1508,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Predastore
|
||||
|
||||
+79
-139
@@ -12,15 +12,13 @@ on:
|
||||
- go.sum
|
||||
- Makefile
|
||||
- tests/e2e/**
|
||||
- scripts/images/**
|
||||
- scripts/build-system-image.sh
|
||||
- .github/workflows/e2e.yml
|
||||
workflow_dispatch:
|
||||
inputs:
|
||||
suites:
|
||||
description: "Space-separated Make targets to run (e.g. 'e2e-single e2e-multinode e2e-cert e2e-lb e2e-eks e2e-iam e2e-rds')"
|
||||
required: false
|
||||
default: "e2e-single e2e-multinode e2e-cert e2e-iam"
|
||||
default: "e2e-single e2e-multinode e2e-cert e2e-lb e2e-iam"
|
||||
test_run:
|
||||
description: "Optional regex for -test.run (e.g. '^TestMultinodeIPSec$'). Empty = run all tests in suite."
|
||||
required: false
|
||||
@@ -33,75 +31,86 @@ permissions:
|
||||
|
||||
env:
|
||||
GIT_BRANCH: ${{ github.ref_name }}
|
||||
SUITES: ${{ inputs.suites || 'e2e-single e2e-multinode e2e-cert e2e-lb e2e-iam e2e-partialblock' }}
|
||||
TEST_RUN: ${{ inputs.test_run || '' }}
|
||||
SPINIFEX_OTLP_ENDPOINT: ${{ vars.SPINIFEX_OTLP_ENDPOINT || 'http://192.168.1.13:8200' }}
|
||||
SPINIFEX_OTEL_ATTRS: ci.run_id=${{ github.run_id }},ci.workflow=${{ github.workflow }}
|
||||
|
||||
jobs:
|
||||
# A push runs the base set plus the suites its changed paths map to in
|
||||
# suite-sets.sh; a dispatch runs its suites input, or the base set if empty.
|
||||
# Budgets come from the suites' own timeouts there.
|
||||
# The inner timeout must fire before the outer one. `go test` reports where it
|
||||
# stopped and the suite's own failure; a job cancelled on timeout reports only
|
||||
# "The operation was canceled" and takes the artifacts with it.
|
||||
#
|
||||
# A job budget cannot be a literal here because the suite set is a dispatch
|
||||
# input: storagefault asks for 90m and diskperf for 180m, both eligible on one
|
||||
# node, so any fixed number is either wrong for them or wasteful for everyone
|
||||
# else. Resolving it from suite-sets.sh keeps the one source of truth and
|
||||
# leaves the default push set at exactly the 60m it has always had.
|
||||
plan:
|
||||
name: Select Suites
|
||||
name: Resolve Job Budgets
|
||||
runs-on: ubuntu-26.04
|
||||
timeout-minutes: 5
|
||||
outputs:
|
||||
single_legs: ${{ steps.select.outputs.single_legs }}
|
||||
multi_suites: ${{ steps.select.outputs.multi_suites }}
|
||||
multi_timeout: ${{ steps.select.outputs.multi_timeout }}
|
||||
single_timeout: ${{ steps.budgets.outputs.single_timeout }}
|
||||
multi_timeout: ${{ steps.budgets.outputs.multi_timeout }}
|
||||
steps:
|
||||
# Full history without file contents: the diff needs both commits' trees.
|
||||
- uses: actions/checkout@v7.0.1
|
||||
with:
|
||||
path: spinifex
|
||||
fetch-depth: 0
|
||||
filter: blob:none
|
||||
sparse-checkout: .github/scripts
|
||||
|
||||
- name: Select suites
|
||||
id: select
|
||||
working-directory: spinifex
|
||||
- name: Resolve Budgets
|
||||
id: budgets
|
||||
env:
|
||||
EVENT: ${{ github.event_name }}
|
||||
BEFORE: ${{ github.event.before }}
|
||||
IN_SUITES: ${{ inputs.suites }}
|
||||
# Provision, install and teardown around the suites, plus the headroom
|
||||
# a multi-suite job needs for whatever ran before the one that hangs.
|
||||
SETUP_ALLOWANCE_MIN: 15
|
||||
# What the job has always had. Kept as a floor so resolving the budget
|
||||
# can only ever raise it, never quietly shorten an existing run.
|
||||
FLOOR_MIN: 60
|
||||
run: |
|
||||
set -euo pipefail
|
||||
SELECT=.github/scripts/e2e-select.sh
|
||||
if [ "${EVENT}" = push ]; then
|
||||
bash "${SELECT}" push "${BEFORE}" "${GITHUB_SHA}" | tee -a "$GITHUB_OUTPUT"
|
||||
else
|
||||
bash "${SELECT}" suites "${IN_SUITES}" | tee -a "$GITHUB_OUTPUT"
|
||||
fi
|
||||
. "${GITHUB_WORKSPACE}/spinifex/.github/scripts/suite-sets.sh"
|
||||
# Only max() is a sound bound: the suites share one job, so a sum would
|
||||
# demand hours for a set that has never taken them. Max is what makes
|
||||
# the hanging suite panic on its own timeout first, which is the point.
|
||||
budget() {
|
||||
local want="${FLOOR_MIN}" s t
|
||||
for s in $(echo "${SUITES}" | sed 's/e2e-//g' | tr ' ' '\n' | grep -xE "$1"); do
|
||||
t="$(e2e_suite_timeout "${s}")"
|
||||
t=$(( ${t%m} + SETUP_ALLOWANCE_MIN ))
|
||||
[ "${t}" -gt "${want}" ] && want="${t}"
|
||||
done
|
||||
echo "${want}"
|
||||
}
|
||||
single=$(budget "${E2E_SUITES_SINGLE_RE}")
|
||||
multi=$(budget "${E2E_SUITES_MULTI_RE}")
|
||||
echo "single_timeout=${single}" >> "$GITHUB_OUTPUT"
|
||||
echo "multi_timeout=${multi}" >> "$GITHUB_OUTPUT"
|
||||
echo "suites='${SUITES}' -> single ${single}m, multi ${multi}m"
|
||||
|
||||
e2e_single:
|
||||
# The base leg keeps the job and artifact names it always had: mulga's
|
||||
# registry intake finds a run's results by them.
|
||||
name: ${{ matrix.leg == 'base' && 'Single-Node E2E' || format('Single-Node E2E ({0})', matrix.leg) }}
|
||||
# `base` runs on ci-single; `service`, on ci-pseudo, carries the matched
|
||||
# single-node suites so a long one stays off the base leg's critical path.
|
||||
if: needs.plan.outputs.single_legs != '[]'
|
||||
name: Single-Node E2E
|
||||
# The dispatch check gates on exactly the suites this job's Run step
|
||||
# allow-lists, i.e. E2E_SUITES_SINGLE in .github/scripts/suite-sets.sh —
|
||||
# an `if:` expression cannot source that file, so this one list is
|
||||
# hand-mirrored and must be updated with it. Gating on a suite the job
|
||||
# won't run — lb, which only executes in the multinode job — would
|
||||
# provision a tofu VM and pay the Spinifex install only to run nothing and
|
||||
# tear down. Push triggers keep running via the null-check on
|
||||
# inputs.suites.
|
||||
if: github.event_name != 'workflow_dispatch' || contains(github.event.inputs.suites, 'e2e-single') || contains(github.event.inputs.suites, 'e2e-iam') || contains(github.event.inputs.suites, 'e2e-cert') || contains(github.event.inputs.suites, 'e2e-eks') || contains(github.event.inputs.suites, 'e2e-ecs') || contains(github.event.inputs.suites, 'e2e-storagegrowth') || contains(github.event.inputs.suites, 'e2e-partialblock') || contains(github.event.inputs.suites, 'e2e-rds')
|
||||
needs: plan
|
||||
runs-on: [self-hosted, "${{ matrix.runner }}"]
|
||||
timeout-minutes: ${{ matrix.timeout }}
|
||||
strategy:
|
||||
fail-fast: false
|
||||
matrix:
|
||||
include: ${{ fromJSON(needs.plan.outputs.single_legs) }}
|
||||
runs-on: [self-hosted, ci-single]
|
||||
timeout-minutes: ${{ fromJSON(needs.plan.outputs.single_timeout) }}
|
||||
env:
|
||||
SUITES: ${{ matrix.suites }}
|
||||
# Both legs can run on banksia at once, so every host path carries the leg.
|
||||
TOFU_LOG: /tmp/e2e-tofu-single-${{ github.run_id }}-${{ matrix.leg }}.log
|
||||
ARTIFACT_DIR: /tmp/spinifex-e2e-artifacts-${{ github.run_id }}-${{ matrix.leg }}
|
||||
# The base leg keeps the cell name it always had, so its history in
|
||||
# Elastic and telemetry continues; any other leg appends its name.
|
||||
E2E_CELL: ${{ matrix.leg == 'base' && 'single' || format('single-{0}', matrix.leg) }}
|
||||
TOFU_LOG: /tmp/e2e-tofu-single-${{ github.run_id }}.log
|
||||
ARTIFACT_DIR: /tmp/spinifex-e2e-artifacts-${{ github.run_id }}
|
||||
# Job-level env wins over the workflow-level default. This job and
|
||||
# e2e_multi share one run id and both name their hosts node1/node2/node3,
|
||||
# so without ci.cell their telemetry collapses into one bucket. This job
|
||||
# is where "Bootstrap via Binary Install" bakes SPINIFEX_OTEL_ATTRS into
|
||||
# the VM's telemetry.env, so the label has to be set here.
|
||||
SPINIFEX_OTEL_ATTRS: ci.run_id=${{ github.run_id }},ci.workflow=${{ github.workflow }},ci.cell=${{ matrix.leg == 'base' && 'single' || format('single-{0}', matrix.leg) }}
|
||||
SPINIFEX_OTEL_ATTRS: ci.run_id=${{ github.run_id }},ci.workflow=${{ github.workflow }},ci.cell=single
|
||||
steps:
|
||||
- name: Reap Orphan Procs
|
||||
run: |
|
||||
@@ -171,7 +180,7 @@ jobs:
|
||||
- name: Resolve CI job id
|
||||
uses: $/.github/actions/otel-job-id
|
||||
with:
|
||||
job-name: ${{ matrix.leg == 'base' && 'Single-Node E2E' || format('Single-Node E2E ({0})', matrix.leg) }}
|
||||
job-name: Single-Node E2E
|
||||
|
||||
- name: Build Install Artifacts
|
||||
working-directory: mulga/scripts/tofu-cluster
|
||||
@@ -215,12 +224,6 @@ jobs:
|
||||
echo "wan_ip=${WAN_IP}" >> "$GITHUB_OUTPUT"
|
||||
echo "ssh_key=${SSH_KEY}" >> "$GITHUB_OUTPUT"
|
||||
|
||||
# build-system-image.sh reuses a raw image under ten minutes old from
|
||||
# its build directory, and every runner on banksia shares the default
|
||||
# one in /tmp, so a job could test another commit's guest agent.
|
||||
- name: Use a per-job image build directory
|
||||
run: echo "SYSTEM_IMAGE_BUILD_DIR=${RUNNER_TEMP}/system-image-build" >> "$GITHUB_ENV"
|
||||
|
||||
# The eks suite's CreateCluster boots a control-plane VM from the
|
||||
# eks-node AMI, so that AMI must exist in the VM's image store before the
|
||||
# suite runs. build-system-image.sh customizes the image with libguestfs
|
||||
@@ -244,7 +247,7 @@ jobs:
|
||||
group_open "make build-eks-node-image"
|
||||
make build-eks-node-image
|
||||
group_close
|
||||
ok "raw: $(du -h "${SYSTEM_IMAGE_BUILD_DIR}/eks-node-image-build/eks-node-alpine.raw" | cut -f1)"
|
||||
ok "raw: $(du -h /tmp/eks-node-image-build/eks-node-alpine.raw | cut -f1)"
|
||||
|
||||
- name: Import eks-node AMI on VM
|
||||
if: contains(env.SUITES, 'e2e-eks')
|
||||
@@ -256,16 +259,13 @@ jobs:
|
||||
SSH_OPTS=(-o StrictHostKeyChecking=no -i "${{ steps.vm.outputs.ssh_key }}")
|
||||
VM=tf-user@${{ steps.vm.outputs.wan_ip }}
|
||||
banner "import eks-node AMI to VM image store"
|
||||
# The VM's /tmp is a tmpfs and scp writes each raw at its full size,
|
||||
# so every import removes its file before the next image arrives.
|
||||
scp "${SSH_OPTS[@]}" \
|
||||
"${SYSTEM_IMAGE_BUILD_DIR}/eks-node-image-build/eks-node-alpine.raw" \
|
||||
/tmp/eks-node-image-build/eks-node-alpine.raw \
|
||||
"$VM":/tmp/eks-node-alpine.raw
|
||||
ssh "${SSH_OPTS[@]}" "$VM" \
|
||||
"spx admin images import --file /tmp/eks-node-alpine.raw \
|
||||
--distro alpine --version ${ALPINE_VERSION} --arch x86_64 \
|
||||
--boot-mode bios --ami-name ${AMI_NAME} --tag ${SYSTEM_TAG}; \
|
||||
rc=\$?; rm -f /tmp/eks-node-alpine.raw; exit \$rc" >/dev/null 2>&1
|
||||
--boot-mode bios --ami-name ${AMI_NAME} --tag ${SYSTEM_TAG}" >/dev/null 2>&1
|
||||
ok "eks-node AMI imported"
|
||||
|
||||
# The eks data-plane subtest (TestEKS/KubectlGetNodes) drives the full
|
||||
@@ -299,40 +299,6 @@ jobs:
|
||||
EOF
|
||||
ok "kubectl installed on VM"
|
||||
|
||||
# The ecs suite boots its container instance from the ecs-node AMI, found
|
||||
# by its managed-by tag, so the tag matters. The build directory is named
|
||||
# for the manifest's IMAGE_NAME (ecs-agent), not for the make target.
|
||||
- name: Build ecs-node AMI (libguestfs, runner)
|
||||
if: contains(env.SUITES, 'e2e-ecs')
|
||||
working-directory: spinifex
|
||||
run: |
|
||||
set -euo pipefail
|
||||
# direct backend = no libvirt. Set here rather than in the script:
|
||||
# it has to be in the environment the build itself runs in.
|
||||
export LIBGUESTFS_BACKEND=direct
|
||||
"${GITHUB_WORKSPACE}/spinifex/.github/scripts/ensure-libguestfs.sh"
|
||||
make build-ecs-node-image
|
||||
|
||||
- name: Import ecs-node AMI on VM
|
||||
if: contains(env.SUITES, 'e2e-ecs')
|
||||
run: |
|
||||
set -euo pipefail
|
||||
source "${GITHUB_WORKSPACE}/spinifex/.github/scripts/ci-fmt.sh"
|
||||
eval "$(grep -E '^(ALPINE_VERSION|AMI_NAME|SYSTEM_TAG)=' \
|
||||
spinifex/scripts/images/ecs-agent/manifest.conf)"
|
||||
SSH_OPTS=(-o StrictHostKeyChecking=no -i "${{ steps.vm.outputs.ssh_key }}")
|
||||
VM=tf-user@${{ steps.vm.outputs.wan_ip }}
|
||||
banner "import ecs-node AMI to VM image store"
|
||||
scp "${SSH_OPTS[@]}" \
|
||||
"${SYSTEM_IMAGE_BUILD_DIR}/ecs-agent-image-build/ecs-agent-alpine.raw" \
|
||||
"$VM":/tmp/ecs-agent-alpine.raw
|
||||
ssh "${SSH_OPTS[@]}" "$VM" \
|
||||
"spx admin images import --file /tmp/ecs-agent-alpine.raw \
|
||||
--distro alpine --version ${ALPINE_VERSION} --arch x86_64 \
|
||||
--boot-mode bios --ami-name ${AMI_NAME} --tag ${SYSTEM_TAG}; \
|
||||
rc=\$?; rm -f /tmp/ecs-agent-alpine.raw; exit \$rc" >/dev/null 2>&1
|
||||
ok "ecs-node AMI imported"
|
||||
|
||||
# CreateDBInstance boots the DB VM from an RDS AMI, so both engines' images
|
||||
# must be in the VM's image store before the suite runs — same split as the
|
||||
# eks-node pair above: build on the runner (needs source + network), import
|
||||
@@ -355,7 +321,7 @@ jobs:
|
||||
group_open "make build-${IMAGE}-image"
|
||||
make "build-${IMAGE}-image"
|
||||
group_close
|
||||
ok "raw: $(du -h "${SYSTEM_IMAGE_BUILD_DIR}/${IMAGE}-image-build/${IMAGE}-alpine.raw" | cut -f1)"
|
||||
ok "raw: $(du -h "/tmp/${IMAGE}-image-build/${IMAGE}-alpine.raw" | cut -f1)"
|
||||
done
|
||||
|
||||
- name: Import RDS AMIs on VM
|
||||
@@ -377,13 +343,12 @@ jobs:
|
||||
for tag in ${SYSTEM_TAG}; do TAG_ARGS="${TAG_ARGS} --tag ${tag}"; done
|
||||
banner "import ${IMAGE} AMI to VM image store"
|
||||
scp "${SSH_OPTS[@]}" \
|
||||
"${SYSTEM_IMAGE_BUILD_DIR}/${IMAGE}-image-build/${IMAGE}-alpine.raw" \
|
||||
"/tmp/${IMAGE}-image-build/${IMAGE}-alpine.raw" \
|
||||
"$VM":"/tmp/${IMAGE}-alpine.raw"
|
||||
ssh "${SSH_OPTS[@]}" "$VM" \
|
||||
"spx admin images import --file /tmp/${IMAGE}-alpine.raw \
|
||||
--distro alpine --version ${ALPINE_VERSION} --arch x86_64 \
|
||||
--boot-mode bios --ami-name ${AMI_NAME}${TAG_ARGS}; \
|
||||
rc=\$?; rm -f /tmp/${IMAGE}-alpine.raw; exit \$rc" >/dev/null 2>&1
|
||||
--boot-mode bios --ami-name ${AMI_NAME}${TAG_ARGS}" >/dev/null 2>&1
|
||||
ok "${IMAGE} AMI imported"
|
||||
done
|
||||
|
||||
@@ -561,7 +526,7 @@ jobs:
|
||||
uses: $/.github/actions/junit-to-es
|
||||
with:
|
||||
junit-glob: ${{ env.ARTIFACT_DIR }}/junit-*.xml
|
||||
cell: ${{ env.E2E_CELL }}
|
||||
cell: single
|
||||
|
||||
- name: Analyze e2e failures
|
||||
if: always()
|
||||
@@ -571,17 +536,10 @@ jobs:
|
||||
log-dir: ${{ env.ARTIFACT_DIR }}
|
||||
title: "E2E failure analysis"
|
||||
|
||||
# The WAN capture is most of the artifact and only matters when a run
|
||||
# failed. Anything short of a clean pass (suite rc, a failed setup or
|
||||
# collect step, a missing rc) keeps it.
|
||||
- name: Drop the packet capture from a passing single-node run
|
||||
if: always() && steps.e2e.outputs.rc == '0' && job.status == 'success'
|
||||
run: rm -f "${ARTIFACT_DIR}"/net-diag/wan.pcap
|
||||
|
||||
- uses: actions/upload-artifact@v7.0.2
|
||||
if: always()
|
||||
with:
|
||||
name: ${{ matrix.leg == 'base' && format('e2e-artifacts-{0}', github.run_id) || format('e2e-artifacts-{0}-{1}', github.run_id, matrix.leg) }}
|
||||
name: e2e-artifacts-${{ github.run_id }}
|
||||
path: ${{ env.ARTIFACT_DIR }}
|
||||
if-no-files-found: warn
|
||||
retention-days: 14
|
||||
@@ -624,7 +582,7 @@ jobs:
|
||||
if: always()
|
||||
uses: actions/upload-artifact@v7.0.2
|
||||
with:
|
||||
name: ${{ matrix.leg == 'base' && format('tofu-log-single-{0}', github.run_id) || format('tofu-log-single-{0}-{1}', github.run_id, matrix.leg) }}
|
||||
name: tofu-log-single-${{ github.run_id }}
|
||||
path: ${{ env.TOFU_LOG }}
|
||||
if-no-files-found: ignore
|
||||
retention-days: 14
|
||||
@@ -640,13 +598,15 @@ jobs:
|
||||
|
||||
e2e_multi:
|
||||
name: Multi-Node E2E
|
||||
# Runs when the plan job selected any multi-node suite, and only then.
|
||||
if: needs.plan.outputs.multi_suites != ''
|
||||
# The dispatch check gates on exactly the suites this job's Run step
|
||||
# allow-lists (multinode|lb|cert) so lb — which only runs here — still
|
||||
# provisions the 3-node env when requested on its own, and the env stays
|
||||
# unprovisioned when none of them are asked for.
|
||||
if: github.event_name != 'workflow_dispatch' || contains(github.event.inputs.suites, 'e2e-multinode') || contains(github.event.inputs.suites, 'e2e-lb') || contains(github.event.inputs.suites, 'e2e-cert')
|
||||
needs: plan
|
||||
runs-on: [self-hosted, ci-multi]
|
||||
timeout-minutes: ${{ fromJSON(needs.plan.outputs.multi_timeout) }}
|
||||
env:
|
||||
SUITES: ${{ needs.plan.outputs.multi_suites }}
|
||||
TOFU_LOG: /tmp/e2e-tofu-multi-${{ github.run_id }}.log
|
||||
ARTIFACT_DIR: /tmp/spinifex-e2e-multi-artifacts-${{ github.run_id }}
|
||||
# See the matching note on e2e_single: ci.cell is what separates this
|
||||
@@ -1030,13 +990,6 @@ jobs:
|
||||
log-dir: ${{ env.ARTIFACT_DIR }}
|
||||
title: "Multi-Node E2E failure analysis"
|
||||
|
||||
# The WAN captures are 90% of the artifact and only matter when a run
|
||||
# failed. Anything short of a clean pass (suite rc, a failed setup or
|
||||
# collect step, a missing rc) keeps them.
|
||||
- name: Drop the packet captures from a passing multi-node run
|
||||
if: always() && steps.e2e.outputs.rc == '0' && job.status == 'success'
|
||||
run: rm -f "${ARTIFACT_DIR}"/net-diag-node*/wan.pcap
|
||||
|
||||
- uses: actions/upload-artifact@v7.0.2
|
||||
if: always()
|
||||
with:
|
||||
@@ -1103,10 +1056,7 @@ jobs:
|
||||
e2e_status:
|
||||
name: Report E2E Status
|
||||
if: always()
|
||||
# plan is a dependency because a failed plan skips both e2e jobs, and a
|
||||
# skipped job counts as passed below: without it a broken selection would
|
||||
# report green having run nothing.
|
||||
needs: [plan, e2e_single, e2e_multi]
|
||||
needs: [e2e_single, e2e_multi]
|
||||
runs-on: ubuntu-26.04
|
||||
timeout-minutes: 5
|
||||
steps:
|
||||
@@ -1115,16 +1065,15 @@ jobs:
|
||||
GH_TOKEN: ${{ github.token }}
|
||||
REPORT_SHA: ${{ github.sha }}
|
||||
run: |
|
||||
plan_ok=true; single_ok=true; multi_ok=true
|
||||
[ "${{ needs.plan.result }}" = success ] || plan_ok=false
|
||||
single_ok=true; multi_ok=true
|
||||
case "${{ needs.e2e_single.result }}" in success|skipped) ;; *) single_ok=false ;; esac
|
||||
case "${{ needs.e2e_multi.result }}" in success|skipped) ;; *) multi_ok=false ;; esac
|
||||
if $plan_ok && $single_ok && $multi_ok; then
|
||||
if $single_ok && $multi_ok; then
|
||||
STATE="success"
|
||||
DESC="E2E suites passed"
|
||||
else
|
||||
STATE="failure"
|
||||
DESC="E2E failed: plan=${{ needs.plan.result }} single=${{ needs.e2e_single.result }} multi=${{ needs.e2e_multi.result }}"
|
||||
DESC="E2E failed: single=${{ needs.e2e_single.result }} multi=${{ needs.e2e_multi.result }}"
|
||||
fi
|
||||
gh api "repos/${{ github.repository }}/statuses/${REPORT_SHA}" \
|
||||
-f state="$STATE" \
|
||||
@@ -1141,7 +1090,7 @@ jobs:
|
||||
# status-report check.
|
||||
e2e_es_summary:
|
||||
name: ES CI Run Summary
|
||||
needs: [plan, e2e_single, e2e_multi]
|
||||
needs: [e2e_single, e2e_multi]
|
||||
if: always()
|
||||
runs-on: [self-hosted, ci-single]
|
||||
timeout-minutes: 5
|
||||
@@ -1153,11 +1102,6 @@ jobs:
|
||||
# this host, so no credentials are needed for the POST.
|
||||
MULGA_ES_SINK: ${{ vars.MULGA_ES_SINK || '192.168.1.13:9200' }}
|
||||
steps:
|
||||
# The workspace outlives the run on a self-hosted runner, and the status
|
||||
# files are collected recursively, so an earlier run's would count here.
|
||||
- name: Clear previous summary downloads
|
||||
run: rm -rf "${GITHUB_WORKSPACE}/es-summary-single" "${GITHUB_WORKSPACE}/es-summary-multi"
|
||||
|
||||
# Either leg may not have run (e.g. a workflow_dispatch that only
|
||||
# requested single-node suites), so a missing artifact is tolerated —
|
||||
# the aggregation script below treats an absent suite-status.txt as
|
||||
@@ -1166,7 +1110,7 @@ jobs:
|
||||
uses: actions/download-artifact@v8.0.2
|
||||
continue-on-error: true
|
||||
with:
|
||||
pattern: e2e-artifacts-${{ github.run_id }}*
|
||||
name: e2e-artifacts-${{ github.run_id }}
|
||||
path: ${{ github.workspace }}/es-summary-single
|
||||
|
||||
- name: Download multi-node artifacts
|
||||
@@ -1186,7 +1130,6 @@ jobs:
|
||||
BRANCH: ${{ github.head_ref || github.ref_name }}
|
||||
SHA: ${{ github.sha }}
|
||||
SERVER_URL: ${{ github.server_url }}
|
||||
PLAN_RESULT: ${{ needs.plan.result }}
|
||||
SINGLE_RESULT: ${{ needs.e2e_single.result }}
|
||||
MULTI_RESULT: ${{ needs.e2e_multi.result }}
|
||||
SINGLE_DIR: ${{ github.workspace }}/es-summary-single
|
||||
@@ -1194,8 +1137,7 @@ jobs:
|
||||
run: |
|
||||
set -euo pipefail
|
||||
|
||||
# One subdirectory per single-node leg's artifact.
|
||||
SINGLE_STATUS=$(find "${SINGLE_DIR}" -name suite-status.txt 2>/dev/null | sort || true)
|
||||
SINGLE_STATUS="${SINGLE_DIR}/suite-status.txt"
|
||||
MULTI_STATUS="${MULTI_DIR}/suite-status.txt"
|
||||
|
||||
# Collect "<suite>: FAILED" lines from both legs' suite-status.txt
|
||||
@@ -1204,9 +1146,7 @@ jobs:
|
||||
# rather than a bash array so this stays correct under `set -u`
|
||||
# even when nothing failed.
|
||||
FAILED_SUITES=""
|
||||
# SINGLE_STATUS is a newline-separated list, so it is split here.
|
||||
# shellcheck disable=SC2086
|
||||
for f in ${SINGLE_STATUS} "$MULTI_STATUS"; do
|
||||
for f in "$SINGLE_STATUS" "$MULTI_STATUS"; do
|
||||
[ -f "$f" ] || continue
|
||||
while IFS= read -r line; do
|
||||
case "$line" in
|
||||
@@ -1226,7 +1166,7 @@ jobs:
|
||||
case "$MULTI_RESULT" in success | skipped) ;; *) multi_ok=false ;; esac
|
||||
|
||||
STATE=success
|
||||
if [ "$PLAN_RESULT" != success ] || [ "$single_ok" != true ] || [ "$multi_ok" != true ] || [ "$FAILED_COUNT" -gt 0 ]; then
|
||||
if [ "$single_ok" != true ] || [ "$multi_ok" != true ] || [ "$FAILED_COUNT" -gt 0 ]; then
|
||||
STATE=failure
|
||||
fi
|
||||
|
||||
|
||||
@@ -113,7 +113,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ env.GIT_BRANCH }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- uses: actions/checkout@v7.0.1
|
||||
|
||||
@@ -84,7 +84,7 @@ jobs:
|
||||
|
||||
- name: Resolve sibling release tags from go.mod
|
||||
id: siblings
|
||||
run: bash spinifex/.github/scripts/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
run: bash spinifex/.github/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
|
||||
- name: Checkout Viperblock
|
||||
uses: actions/checkout@v7.0.1
|
||||
|
||||
@@ -18,7 +18,7 @@ jobs:
|
||||
|
||||
- name: Resolve sibling release tags from go.mod
|
||||
id: siblings
|
||||
run: bash spinifex/.github/scripts/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
run: bash spinifex/.github/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
|
||||
- name: Checkout Predastore
|
||||
uses: actions/checkout@v7.0.1
|
||||
@@ -57,8 +57,8 @@ jobs:
|
||||
go-version-file: spinifex/go.mod
|
||||
cache: false
|
||||
|
||||
- name: Build microVM artifacts (lb-agent + kernel + initramfs + qboot + SQS store and worker images)
|
||||
run: cd spinifex && make build-microvm-image build-sqs-worker-microvm-image
|
||||
- name: Build microVM artifacts (lb-agent + kernel + initramfs + qboot)
|
||||
run: cd spinifex && make build-microvm-image
|
||||
|
||||
- name: Build amd64 Tarball
|
||||
env:
|
||||
@@ -85,7 +85,7 @@ jobs:
|
||||
|
||||
- name: Resolve sibling release tags from go.mod
|
||||
id: siblings
|
||||
run: bash spinifex/.github/scripts/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
run: bash spinifex/.github/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
|
||||
- name: Checkout Predastore
|
||||
uses: actions/checkout@v7.0.1
|
||||
@@ -335,7 +335,7 @@ jobs:
|
||||
|
||||
- name: Resolve sibling release tags from go.mod
|
||||
id: siblings
|
||||
run: bash spinifex/.github/scripts/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
run: bash spinifex/.github/resolve-release-sibling-refs.sh spinifex/go.mod
|
||||
|
||||
- name: Checkout Predastore
|
||||
uses: actions/checkout@v7.0.1
|
||||
|
||||
@@ -34,7 +34,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.head_ref || github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Viperblock
|
||||
|
||||
@@ -20,12 +20,6 @@ on:
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
# A newer push to the same pull request supersedes its run. Pushes and dispatches
|
||||
# get a group of their own, so every commit on dev and main keeps its result.
|
||||
concurrency:
|
||||
group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && format('pr-{0}', github.event.pull_request.number) || github.run_id }}
|
||||
cancel-in-progress: true
|
||||
|
||||
jobs:
|
||||
build_and_test:
|
||||
name: Build and Test
|
||||
@@ -42,7 +36,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.head_ref || github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Viperblock
|
||||
@@ -105,13 +99,6 @@ jobs:
|
||||
key: gobuild-${{ github.job }}-${{ github.run_id }}
|
||||
restore-keys: gobuild-${{ github.job }}-
|
||||
|
||||
# Go refreshes an entry's mtime on use only when it is over an hour old, so
|
||||
# restored entries are backdated past that before any Go command runs and
|
||||
# the trim below can tell this run's working set from the last run's.
|
||||
- name: Mark Go build cache
|
||||
if: ${{ github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main' }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh mark
|
||||
|
||||
- name: Setup Go Workspace
|
||||
run: go work init ./spinifex ./viperblock ./predastore ./northstar ./bluebottle
|
||||
|
||||
@@ -132,6 +119,9 @@ jobs:
|
||||
- name: Build Spinifex
|
||||
run: make -C spinifex build
|
||||
|
||||
- name: Manifest check + drift lint
|
||||
run: make -C spinifex manifest-check manifest-lint
|
||||
|
||||
- name: Run Tests with Coverage
|
||||
env:
|
||||
SPINIFEX_REQUIRE_CONFORMANCE_TOOLS: "1"
|
||||
@@ -144,12 +134,19 @@ jobs:
|
||||
- name: Check Diff Coverage
|
||||
run: cd spinifex && scripts/diff-coverage.sh coverage.out --base HEAD~1
|
||||
|
||||
# Keeps what this run read or wrote and drops the rest, so the saved entry
|
||||
# is one run's working set however often dev is pushed. A red run may have
|
||||
# stopped early and used only part of the set, so it saves untrimmed.
|
||||
# Compiled output stays valid whatever the tests said, and skipping the
|
||||
# save on a red run would leave the next one rebuilding it, so this is
|
||||
# gated on the branch rather than on success.
|
||||
# Go bumps an entry's mtime when a run uses it, so this run's working set
|
||||
# survives and objects superseded by later commits do not. Runs after the
|
||||
# work, so it costs this job nothing and only shrinks what the next reads.
|
||||
- name: Trim Go build cache
|
||||
if: ${{ success() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh trim
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: |
|
||||
C="$(go env GOCACHE)"
|
||||
before=$(du -sm "$C" | cut -f1)
|
||||
find "$C" -type f -mmin +240 -delete
|
||||
echo "trimmed ${before}MB -> $(du -sm "$C" | cut -f1)MB"
|
||||
|
||||
- name: Save Go build cache
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
@@ -160,7 +157,6 @@ jobs:
|
||||
|
||||
race_detection:
|
||||
name: Race Detection
|
||||
if: ${{ github.event_name != 'pull_request' }}
|
||||
runs-on: ubuntu-26.04
|
||||
|
||||
steps:
|
||||
@@ -173,7 +169,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.head_ref || github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Viperblock
|
||||
@@ -236,25 +232,22 @@ jobs:
|
||||
key: gobuild-${{ github.job }}-${{ github.run_id }}
|
||||
restore-keys: gobuild-${{ github.job }}-
|
||||
|
||||
# Go refreshes an entry's mtime on use only when it is over an hour old, so
|
||||
# restored entries are backdated past that before any Go command runs and
|
||||
# the trim below can tell this run's working set from the last run's.
|
||||
- name: Mark Go build cache
|
||||
if: ${{ github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main' }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh mark
|
||||
|
||||
- name: Setup Go Workspace
|
||||
run: go work init ./spinifex ./viperblock ./predastore ./northstar ./bluebottle
|
||||
|
||||
- name: Run Tests with Race Detector
|
||||
run: make -C spinifex test-race GOTESTFLAGS=-count=1
|
||||
|
||||
# Keeps what this run read or wrote and drops the rest, so the saved entry
|
||||
# is one run's working set however often dev is pushed. A red run may have
|
||||
# stopped early and used only part of the set, so it saves untrimmed.
|
||||
# Go bumps an entry's mtime when a run uses it, so this run's working set
|
||||
# survives and objects superseded by later commits do not. Runs after the
|
||||
# work, so it costs this job nothing and only shrinks what the next reads.
|
||||
- name: Trim Go build cache
|
||||
if: ${{ success() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh trim
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: |
|
||||
C="$(go env GOCACHE)"
|
||||
before=$(du -sm "$C" | cut -f1)
|
||||
find "$C" -type f -mmin +240 -delete
|
||||
echo "trimmed ${before}MB -> $(du -sm "$C" | cut -f1)MB"
|
||||
|
||||
- name: Save Go build cache
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
@@ -278,7 +271,7 @@ jobs:
|
||||
id: siblings
|
||||
env:
|
||||
BRANCH: ${{ github.head_ref || github.ref_name }}
|
||||
run: bash .github/scripts/resolve-sibling-refs.sh
|
||||
run: bash .github/resolve-sibling-refs.sh
|
||||
working-directory: spinifex
|
||||
|
||||
- name: Checkout Viperblock
|
||||
@@ -341,13 +334,6 @@ jobs:
|
||||
key: gobuild-${{ github.job }}-${{ github.run_id }}
|
||||
restore-keys: gobuild-${{ github.job }}-
|
||||
|
||||
# Go refreshes an entry's mtime on use only when it is over an hour old, so
|
||||
# restored entries are backdated past that before any Go command runs and
|
||||
# the trim below can tell this run's working set from the last run's.
|
||||
- name: Mark Go build cache
|
||||
if: ${{ github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main' }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh mark
|
||||
|
||||
- name: Setup Go Workspace
|
||||
run: go work init ./spinifex ./viperblock ./predastore ./northstar ./bluebottle
|
||||
|
||||
@@ -364,16 +350,16 @@ jobs:
|
||||
if: ${{ !cancelled() }}
|
||||
run: make -C spinifex test-integration
|
||||
|
||||
- name: Run Gated Suites
|
||||
if: ${{ !cancelled() }}
|
||||
run: make -C spinifex test-gated GOTESTFLAGS=-count=1
|
||||
|
||||
# Keeps what this run read or wrote and drops the rest, so the saved entry
|
||||
# is one run's working set however often dev is pushed. A red run may have
|
||||
# stopped early and used only part of the set, so it saves untrimmed.
|
||||
# Go bumps an entry's mtime when a run uses it, so this run's working set
|
||||
# survives and objects superseded by later commits do not. Runs after the
|
||||
# work, so it costs this job nothing and only shrinks what the next reads.
|
||||
- name: Trim Go build cache
|
||||
if: ${{ success() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: spinifex/.github/scripts/go-build-cache.sh trim
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
run: |
|
||||
C="$(go env GOCACHE)"
|
||||
before=$(du -sm "$C" | cut -f1)
|
||||
find "$C" -type f -mmin +240 -delete
|
||||
echo "trimmed ${before}MB -> $(du -sm "$C" | cut -f1)MB"
|
||||
|
||||
- name: Save Go build cache
|
||||
if: ${{ !cancelled() && (github.ref == 'refs/heads/dev' || github.ref == 'refs/heads/main') }}
|
||||
|
||||
@@ -7,11 +7,6 @@ on:
|
||||
- dev
|
||||
paths:
|
||||
- spinifex/runtime/roles/spinifexui/frontend/**
|
||||
- spinifex/runtime/roles/spinifexui/*.go
|
||||
- spinifex/operator/admin/instance_metrics*.go
|
||||
- spinifex/operator/admin/live_streams*.go
|
||||
- spinifex/runtime/roles/qmpcollector/**
|
||||
- contracts/telemetry/**
|
||||
- .github/workflows/ui.yml
|
||||
pull_request:
|
||||
branches:
|
||||
@@ -19,23 +14,12 @@ on:
|
||||
- dev
|
||||
paths:
|
||||
- spinifex/runtime/roles/spinifexui/frontend/**
|
||||
- spinifex/runtime/roles/spinifexui/*.go
|
||||
- spinifex/operator/admin/instance_metrics*.go
|
||||
- spinifex/operator/admin/live_streams*.go
|
||||
- spinifex/runtime/roles/qmpcollector/**
|
||||
- contracts/telemetry/**
|
||||
- .github/workflows/ui.yml
|
||||
workflow_dispatch:
|
||||
|
||||
permissions:
|
||||
contents: read
|
||||
|
||||
# A newer push to the same pull request supersedes its run. Pushes and dispatches
|
||||
# get a group of their own, so every commit on dev and main keeps its result.
|
||||
concurrency:
|
||||
group: ${{ github.workflow }}-${{ github.event_name == 'pull_request' && format('pr-{0}', github.event.pull_request.number) || github.run_id }}
|
||||
cancel-in-progress: true
|
||||
|
||||
jobs:
|
||||
ui_build_and_test:
|
||||
name: UI Build and Test
|
||||
@@ -85,85 +69,3 @@ jobs:
|
||||
echo "$drift"
|
||||
exit 1
|
||||
fi
|
||||
|
||||
# The pinned image lives in one file that scripts/ui-playwright.sh also reads,
|
||||
# so make ui-baselines and the job below draw screenshots in the same image.
|
||||
ui_e2e_image:
|
||||
name: UI End-to-End Image
|
||||
runs-on: ubuntu-26.04
|
||||
outputs:
|
||||
image: ${{ steps.read.outputs.image }}
|
||||
|
||||
steps:
|
||||
- name: Checkout Code
|
||||
uses: actions/checkout@v7.0.1
|
||||
with:
|
||||
sparse-checkout: spinifex/runtime/roles/spinifexui/frontend/playwright.image
|
||||
sparse-checkout-cone-mode: false
|
||||
|
||||
- name: Read Playwright Image
|
||||
id: read
|
||||
run: echo "image=$(tr -d '[:space:]' < spinifex/runtime/roles/spinifexui/frontend/playwright.image)" >> "$GITHUB_OUTPUT"
|
||||
|
||||
# The browser suite: the console against a faked API on Vite dev, plus the
|
||||
# built dist under the production CSP. It runs in Playwright's own image, which
|
||||
# carries the browsers, so screenshots rasterise as they did for the baselines.
|
||||
ui_e2e:
|
||||
name: UI End-to-End
|
||||
needs: ui_e2e_image
|
||||
runs-on: ubuntu-26.04
|
||||
timeout-minutes: 20
|
||||
container:
|
||||
image: ${{ needs.ui_e2e_image.outputs.image }}
|
||||
options: --ipc=host
|
||||
defaults:
|
||||
run:
|
||||
working-directory: spinifex/runtime/roles/spinifexui/frontend
|
||||
|
||||
steps:
|
||||
- name: Checkout Code
|
||||
uses: actions/checkout@v7.0.1
|
||||
with:
|
||||
sparse-checkout: spinifex/runtime/roles/spinifexui/frontend
|
||||
|
||||
- name: Setup pnpm
|
||||
uses: pnpm/action-setup@v6.1.0
|
||||
with:
|
||||
package_json_file: spinifex/runtime/roles/spinifexui/frontend/package.json
|
||||
|
||||
- name: Setup Node.js
|
||||
uses: actions/setup-node@v7.1.0
|
||||
with:
|
||||
cache: "pnpm"
|
||||
cache-dependency-path: spinifex/runtime/roles/spinifexui/frontend/pnpm-lock.yaml
|
||||
node-version-file: spinifex/runtime/roles/spinifexui/frontend/.nvmrc
|
||||
|
||||
- name: Install Dependencies
|
||||
run: pnpm install --frozen-lockfile
|
||||
|
||||
# The image's browsers belong to one Playwright release, so a lockfile
|
||||
# bump without an image bump would otherwise fail on a missing executable.
|
||||
- name: Check Playwright Matches the Image
|
||||
env:
|
||||
IMAGE: ${{ needs.ui_e2e_image.outputs.image }}
|
||||
run: |
|
||||
version=$(pnpm exec playwright --version | awk '{print $2}')
|
||||
case "$IMAGE" in
|
||||
*":v$version-"*) ;;
|
||||
*) echo "::error::Lockfile has Playwright $version but the image is $IMAGE. Update playwright.image."; exit 1 ;;
|
||||
esac
|
||||
|
||||
- name: Build Frontend
|
||||
run: pnpm build
|
||||
|
||||
- name: Run End-to-End Suite
|
||||
run: pnpm e2e
|
||||
|
||||
- name: Upload Screenshot Diffs
|
||||
if: failure()
|
||||
uses: actions/upload-artifact@v7.0.2
|
||||
with:
|
||||
name: ui-e2e-test-results
|
||||
path: spinifex/runtime/roles/spinifexui/frontend/test-results/
|
||||
if-no-files-found: ignore
|
||||
retention-days: 14
|
||||
|
||||
@@ -21,13 +21,6 @@ build/microvm/vmlinuz
|
||||
build/microvm/initramfs.cpio.gz
|
||||
build/microvm/qboot.rom
|
||||
build/microvm/qboot.LICENSE
|
||||
build/microvm/sqs-worker/vmlinuz
|
||||
build/microvm/sqs-worker/initramfs.cpio.gz
|
||||
build/microvm/sqs-worker/.stage.*/
|
||||
build/microvm/lambda-runtime/vmlinuz
|
||||
build/microvm/lambda-runtime/initramfs.cpio.gz
|
||||
build/microvm/lambda-runtime/.stage.*/
|
||||
build/microvm/lambda-testlink-runtime/
|
||||
build/qboot/
|
||||
!spinifex/runtime/roles/spinifexui/frontend/dist/
|
||||
tmp/
|
||||
|
||||
@@ -42,15 +42,11 @@ ifdef QUIET
|
||||
_COVQ = 2>&1 | { grep -Ev '^\s*(ok|PASS|\?|=== RUN|--- PASS:)\s' | grep -v 'coverage: 0\.0%' || true; }; exit $${PIPESTATUS[0]}
|
||||
_RACEQ = 2>&1 | { grep -Ev '^\s*(ok|PASS|\?|=== RUN|--- PASS:)\s' || true; }; exit $${PIPESTATUS[0]}
|
||||
_SECQ = >
|
||||
_HDR = :
|
||||
_RUNSH = sh "$$t" >"$$log" 2>&1 || { echo "-- $$t"; cat "$$log"; exit 1; }
|
||||
else
|
||||
_Q =
|
||||
_COVQ =
|
||||
_RACEQ =
|
||||
_SECQ = 2>&1 | tee
|
||||
_HDR = echo -e
|
||||
_RUNSH = echo "-- $$t"; sh "$$t" || exit 1
|
||||
endif
|
||||
|
||||
build: go_build build-installer build-lb-agent generate-aws-model-coverage
|
||||
@@ -60,19 +56,6 @@ build-ui:
|
||||
@echo -e "\n....Building spinifex-ui frontend...."
|
||||
cd spinifex/runtime/roles/spinifexui/frontend && pnpm build
|
||||
|
||||
# Playwright suite for the console, against Vite dev and the built dist under
|
||||
# the console's CSP. Not part of preflight; needs `pnpm exec playwright install chromium`.
|
||||
test-ui-e2e: build-ui
|
||||
cd spinifex/runtime/roles/spinifexui/frontend && pnpm e2e
|
||||
|
||||
# The same suite inside the pinned Playwright image CI uses (podman or docker),
|
||||
# so screenshots compare as they will on the runner. ui-baselines redraws them.
|
||||
ui-e2e-container:
|
||||
scripts/ui-playwright.sh compare
|
||||
|
||||
ui-baselines:
|
||||
scripts/ui-playwright.sh baselines
|
||||
|
||||
# GO commands
|
||||
VERSION ?= $(shell git describe --tags --always --dirty --exclude '*-dev')
|
||||
COMMIT ?= $(shell git rev-parse --short HEAD)
|
||||
@@ -82,36 +65,6 @@ go_build:
|
||||
@echo -e "\n....Building $(GO_PROJECT_NAME)"
|
||||
go build $(GO_BUILD_MOD) -ldflags "$(LDFLAGS)" -o ./bin/$(GO_PROJECT_NAME) cmd/spinifex/main.go
|
||||
|
||||
# Controlled-test only: builds the tagged daemon and its hard-kill initializer
|
||||
# together, pinning the initializer's digest into that daemon build. The daemon
|
||||
# keeps its symbols so the pin can be read back from the binary.
|
||||
# LAMBDA_IMAGE_DIGEST is the test-link runtime image digest that
|
||||
# build-lambda-testlink-image prints; it is read raw and validated before use.
|
||||
LAMBDA_TESTLINK_DIR := ./bin/lambda-testlink
|
||||
LAMBDA_TESTLINK_PKG := github.com/mulgadc/spinifex/spinifex/runtime/daemon
|
||||
LAMBDA_TESTLINK_PIN := $(LAMBDA_TESTLINK_PKG).lambdaHardKillSHA256
|
||||
LAMBDA_TESTLINK_GW := github.com/mulgadc/spinifex/spinifex/runtime/roles/awsgw
|
||||
LAMBDA_DIGEST_RE := ^[0-9a-f]{64}$$
|
||||
build-lambda-testlink: override export LAMBDA_IMAGE_DIGEST := $(value LAMBDA_IMAGE_DIGEST)
|
||||
build-lambda-testlink:
|
||||
@[[ "$$LAMBDA_IMAGE_DIGEST" =~ $(LAMBDA_DIGEST_RE) ]] || { echo "LAMBDA_IMAGE_DIGEST must be 64 lowercase hex digits"; exit 1; }
|
||||
@host=$$(uname -m); goos=$$(go env GOOS); goarch=$$(go env GOARCH); \
|
||||
test "$$host" = x86_64 && test "$$goos/$$goarch" = linux/amd64 || { echo "the test link supports only an x86_64 host building linux/amd64, not $$host building $$goos/$$goarch"; exit 1; }
|
||||
@rm -f $(LAMBDA_TESTLINK_DIR)/hardkill.so $(LAMBDA_TESTLINK_DIR)/$(GO_PROJECT_NAME)
|
||||
@mkdir -p $(LAMBDA_TESTLINK_DIR).tmp
|
||||
@rm -f $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so $(LAMBDA_TESTLINK_DIR).tmp/$(GO_PROJECT_NAME)
|
||||
cc -std=c17 -O2 -Wall -Wextra -Werror -shared -fPIC -o $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so spinifex/runtime/compute/lambda/testlink/hardkill/hardkill.c -lrt
|
||||
scripts/check-glibc-ceiling.sh 2.34 $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so
|
||||
@digest=$$(sha256sum $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so | cut -d' ' -f1); \
|
||||
test $${#digest} -eq 64 || { echo "initializer digest missing"; exit 1; }; \
|
||||
echo "initializer sha256 $$digest"; \
|
||||
pin="-X $(LAMBDA_TESTLINK_PIN)=$$digest -X $(LAMBDA_TESTLINK_PKG).lambdaImageDigest=$$LAMBDA_IMAGE_DIGEST -X $(LAMBDA_TESTLINK_GW).lambdaImageDigest=$$LAMBDA_IMAGE_DIGEST"; \
|
||||
LAMBDA_TESTLINK_INITIALIZER=$(abspath $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so) go test -count=1 -tags lambdatestlink -ldflags "$$pin" -run '^TestLambda(HardKillPin|ImageDigestPin)$$' ./spinifex/runtime/daemon/ ./spinifex/runtime/roles/awsgw/ && \
|
||||
go build $(GO_BUILD_MOD) -tags lambdatestlink -ldflags "$(subst -s -w ,,$(LDFLAGS)) $$pin" -o $(LAMBDA_TESTLINK_DIR).tmp/$(GO_PROJECT_NAME) cmd/spinifex/main.go
|
||||
@mkdir -p $(LAMBDA_TESTLINK_DIR)
|
||||
mv $(LAMBDA_TESTLINK_DIR).tmp/hardkill.so $(LAMBDA_TESTLINK_DIR).tmp/$(GO_PROJECT_NAME) $(LAMBDA_TESTLINK_DIR)/
|
||||
@rmdir $(LAMBDA_TESTLINK_DIR).tmp
|
||||
|
||||
build-installer:
|
||||
@echo -e "\n....Building spinifex-installer"
|
||||
go build -ldflags "-s -w" -o ./bin/spinifex-installer cmd/installer/main.go
|
||||
@@ -187,65 +140,6 @@ $(MICROVM_ARTIFACTS) &: $(MICROVM_INPUTS)
|
||||
bin/lb-agent:
|
||||
$(MAKE) build-lb-agent
|
||||
|
||||
# The dedicated SQS server, built from the go.mod-pinned module so the image
|
||||
# ships exactly the reviewed version. Static for the musl guest.
|
||||
bin/nats-server: go.mod go.sum
|
||||
@mkdir -p bin
|
||||
GOWORK=off CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -trimpath -ldflags '-s -w' -o $@ github.com/nats-io/nats-server/v2
|
||||
|
||||
# go_local_srcs lists the main module package dirs and files guest package $(1)
|
||||
# links, scanned with the guest build settings; a dir's mtime catches a deleted
|
||||
# or renamed file. A failed scan names a missing prerequisite, so make stops.
|
||||
go_local_srcs = $(shell GOWORK=off CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go list -deps -f '{{if and .Module .Module.Main}}{{.Dir}} {{range .GoFiles}}{{$$.Dir}}/{{.}} {{end}}{{range .EmbedFiles}}{{$$.Dir}}/{{.}} {{end}}{{end}}' $(1) || echo go-list-failed-for-$(1))
|
||||
.SECONDEXPANSION:
|
||||
|
||||
# The SQS microVM image: one guest running its own SQS nats-server and the SQS
|
||||
# worker as separate processes. A separate artifact from the LB image above.
|
||||
SQS_WORKER_MICROVM_OUT_DIR := $(MICROVM_OUT_DIR)/sqs-worker
|
||||
SQS_WORKER_MICROVM_ARTIFACTS := $(SQS_WORKER_MICROVM_OUT_DIR)/vmlinuz $(SQS_WORKER_MICROVM_OUT_DIR)/initramfs.cpio.gz
|
||||
SQS_WORKER_MICROVM_INPUTS := scripts/build-sqs-store-microvm-image.sh scripts/lib/microvm-image.sh scripts/lib/sqs-store-microvm-publish.sh $(SQS_WORKER_MICROVM_OUT_DIR)/init.sh $(SQS_WORKER_MICROVM_OUT_DIR)/inittab $(addprefix $(SQS_WORKER_MICROVM_OUT_DIR)/,sqs-worker-lib.sh sqs-worker-prepare sqs-worker-store sqs-worker-ready sqs-worker-stop) bin/nats-server bin/sqs-worker
|
||||
|
||||
$(SQS_WORKER_MICROVM_ARTIFACTS) &: $(SQS_WORKER_MICROVM_INPUTS)
|
||||
./scripts/build-sqs-store-microvm-image.sh
|
||||
|
||||
bin/sqs-worker: go.mod go.sum $$(call go_local_srcs,./cmd/sqs-worker)
|
||||
@mkdir -p bin
|
||||
GOWORK=off CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -trimpath -ldflags '-s -w' -o $@ ./cmd/sqs-worker
|
||||
|
||||
build-sqs-worker-microvm-image: $(SQS_WORKER_MICROVM_ARTIFACTS) ## Build the SQS microVM kernel and initramfs
|
||||
|
||||
# The Lambda runtime microVM image. Built on demand; not shipped or installed.
|
||||
LAMBDA_RUNTIME_MICROVM_OUT_DIR := $(MICROVM_OUT_DIR)/lambda-runtime
|
||||
LAMBDA_RUNTIME_MICROVM_ARTIFACTS := $(LAMBDA_RUNTIME_MICROVM_OUT_DIR)/vmlinuz $(LAMBDA_RUNTIME_MICROVM_OUT_DIR)/initramfs.cpio.gz
|
||||
LAMBDA_RUNTIME_MICROVM_INPUTS := scripts/build-lambda-runtime-microvm-image.sh scripts/lib/microvm-image.sh $(addprefix $(LAMBDA_RUNTIME_MICROVM_OUT_DIR)/,init.sh inittab run-agent) bin/lambda-agent
|
||||
|
||||
$(LAMBDA_RUNTIME_MICROVM_ARTIFACTS) &: $(LAMBDA_RUNTIME_MICROVM_INPUTS)
|
||||
./scripts/build-lambda-runtime-microvm-image.sh
|
||||
|
||||
bin/lambda-agent: go.mod go.sum $$(call go_local_srcs,./cmd/lambda-agent)
|
||||
@mkdir -p bin
|
||||
GOWORK=off CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -trimpath -ldflags '-s -w' -o $@ ./cmd/lambda-agent
|
||||
|
||||
build-lambda-runtime-microvm-image: $(LAMBDA_RUNTIME_MICROVM_ARTIFACTS) ## Build the Lambda runtime microVM kernel and initramfs (not shipped)
|
||||
GOWORK=off go run ./cmd/lambda-image-agent-tag $(LAMBDA_RUNTIME_MICROVM_OUT_DIR)/initramfs.cpio.gz absent
|
||||
|
||||
# Controlled-test only: a separately built guest whose agent has the
|
||||
# lambdatestlink tag, in its own outputs, never touching bin/lambda-agent or the
|
||||
# default image. It prints the digest to pass as LAMBDA_IMAGE_DIGEST.
|
||||
LAMBDA_TESTLINK_IMAGE_DIR := $(MICROVM_OUT_DIR)/lambda-testlink-runtime
|
||||
build-lambda-testlink-image: ## Build the test-link Lambda guest image and print its digest (not shipped)
|
||||
@host=$$(uname -m); goos=$$(go env GOOS); goarch=$$(go env GOARCH); \
|
||||
test "$$host" = x86_64 && test "$$goos/$$goarch" = linux/amd64 || { echo "the test link supports only an x86_64 host building linux/amd64, not $$host building $$goos/$$goarch"; exit 1; }
|
||||
$(MAKE) --no-print-directory $(MICROVM_OUT_DIR)/qboot.rom
|
||||
@mkdir -p bin $(LAMBDA_TESTLINK_IMAGE_DIR)
|
||||
GOWORK=off GOFLAGS= CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -trimpath -tags lambdatestlink -ldflags '-s -w' -o bin/lambda-agent-testlink ./cmd/lambda-agent
|
||||
./scripts/build-lambda-runtime-microvm-image.sh bin/lambda-agent-testlink $(LAMBDA_TESTLINK_IMAGE_DIR)
|
||||
GOWORK=off go run ./cmd/lambda-image-agent-tag $(LAMBDA_TESTLINK_IMAGE_DIR)/initramfs.cpio.gz present
|
||||
install -m 0644 $(MICROVM_OUT_DIR)/qboot.rom $(LAMBDA_TESTLINK_IMAGE_DIR)/qboot.rom
|
||||
@digest=$$(GOWORK=off go run -tags lambdatestlink ./cmd/lambda-image-digest $(LAMBDA_TESTLINK_IMAGE_DIR)) && \
|
||||
[[ "$$digest" =~ $(LAMBDA_DIGEST_RE) ]] || { echo "image digest not computed"; exit 1; }; \
|
||||
echo "LAMBDA_IMAGE_DIGEST=$$digest"
|
||||
|
||||
# qboot is the microvm firmware. The source tarball ships with each release
|
||||
# because qboot is GPL-2.0.
|
||||
QBOOT_COMMIT := $(shell sed -n 's/^QBOOT_COMMIT="\(.*\)"$$/\1/p' scripts/build-qboot.sh)
|
||||
@@ -257,12 +151,6 @@ $(QBOOT_ARTIFACTS) &: scripts/build-qboot.sh
|
||||
|
||||
build-microvm-image: $(MICROVM_ARTIFACTS) $(QBOOT_ARTIFACTS) ## Build microVM kernel, initramfs and qboot firmware (incremental — skips when up to date)
|
||||
|
||||
install-sqs-worker-microvm: $(SQS_WORKER_MICROVM_ARTIFACTS) ## Install the SQS microVM image to /usr/share/spinifex/microvm/sqs-worker/
|
||||
sudo install -d /usr/share/spinifex/microvm/sqs-worker
|
||||
sudo rm -f /usr/share/spinifex/microvm/sqs-worker/initramfs.cpio.gz
|
||||
sudo install -m 0644 $(SQS_WORKER_MICROVM_OUT_DIR)/vmlinuz /usr/share/spinifex/microvm/sqs-worker/vmlinuz
|
||||
sudo install -m 0644 $(SQS_WORKER_MICROVM_OUT_DIR)/initramfs.cpio.gz /usr/share/spinifex/microvm/sqs-worker/initramfs.cpio.gz
|
||||
|
||||
install-microvm: $(MICROVM_ARTIFACTS) $(QBOOT_ARTIFACTS) ## Install microVM artifacts to /usr/share/spinifex/microvm/
|
||||
sudo install -d /usr/share/spinifex/microvm
|
||||
sudo install -m 0644 $(MICROVM_OUT_DIR)/vmlinuz /usr/share/spinifex/microvm/vmlinuz
|
||||
@@ -270,14 +158,15 @@ install-microvm: $(MICROVM_ARTIFACTS) $(QBOOT_ARTIFACTS) ## Install microVM arti
|
||||
sudo install -m 0644 $(MICROVM_OUT_DIR)/qboot.rom /usr/share/spinifex/microvm/qboot.rom
|
||||
sudo install -m 0644 $(MICROVM_OUT_DIR)/qboot.LICENSE /usr/share/spinifex/microvm/qboot.LICENSE
|
||||
|
||||
# the pre-commit gate: lint, vuln, and the unit and e2e-harness tiers.
|
||||
# the pre-commit gate: manifest checks, lint, vuln, and the unit and e2e-harness tiers.
|
||||
# integration and race tests skipped to keep quick, they run in CI.
|
||||
#
|
||||
# -j2 overlaps the only two heavy gates, lint and test-cover: alone each leaves
|
||||
# most of the machine idle, so pairing them cuts a cold run by about a quarter.
|
||||
# -Otarget keeps each gate's output in one block instead of interleaving them.
|
||||
# manifest-check manifest-lint test-package-check removed, can add back later
|
||||
preflight:
|
||||
@$(MAKE) --no-print-directory -j2 -Otarget QUIET=1 lint govulncheck test-cover diff-coverage test-harness test-build-scripts
|
||||
@$(MAKE) --no-print-directory -j2 -Otarget QUIET=1 lint govulncheck test-cover diff-coverage test-harness test-build-scripts manifest-check manifest-lint
|
||||
@echo -e "\n ✅ Preflight passed — safe to commit."
|
||||
|
||||
# Shell suites + shellcheck for build/scripts/, the systemd-unit helpers that
|
||||
@@ -289,33 +178,22 @@ preflight:
|
||||
# reason: between them they install the software, wire OVN, erase the state and
|
||||
# remove the install, so a shellcheck warning in one is an outage or a wrong
|
||||
# deletion, not a style note.
|
||||
NODE_SCRIPTS := scripts/setup.sh scripts/setup-ovn.sh scripts/install-node.sh scripts/smoke-test.sh scripts/node-reset.sh scripts/uninstall-spx.sh scripts/lib/distro.sh scripts/lib/stage-sqs-store-microvm.sh
|
||||
NODE_SCRIPTS := scripts/setup.sh scripts/setup-ovn.sh scripts/install-node.sh scripts/smoke-test.sh scripts/node-reset.sh scripts/uninstall-spx.sh scripts/lib/distro.sh
|
||||
|
||||
test-build-scripts:
|
||||
@$(_HDR) "\n....Running build/scripts/**/*_test.sh...."
|
||||
@log=$$(mktemp); trap 'rm -f "$$log"' EXIT; for t in $$(find build/scripts -name '*_test.sh' | sort); do \
|
||||
$(_RUNSH); \
|
||||
@echo -e "\n....Running build/scripts/**/*_test.sh...."
|
||||
@for t in $$(find build/scripts -name '*_test.sh' | sort); do \
|
||||
echo "-- $$t"; \
|
||||
sh "$$t" || exit 1; \
|
||||
done
|
||||
@$(_HDR) "\n....Running scripts/*_test.sh, scripts/lib/*_test.sh and scripts/appliance/*_test.sh...."
|
||||
@log=$$(mktemp); trap 'rm -f "$$log"' EXIT; for t in $$( (find scripts -maxdepth 1 -name '*_test.sh'; find scripts/lib scripts/appliance -name '*_test.sh') | sort); do \
|
||||
$(_RUNSH); \
|
||||
@echo -e "\n....Running scripts/*_test.sh and scripts/lib/*_test.sh...."
|
||||
@for t in $$( (find scripts -maxdepth 1 -name '*_test.sh'; find scripts/lib -name '*_test.sh') | sort); do \
|
||||
echo "-- $$t"; \
|
||||
sh "$$t" || exit 1; \
|
||||
done
|
||||
@$(_HDR) "\n....Running scripts/terraform/latitude-spx/*_test.sh...."
|
||||
@log=$$(mktemp); trap 'rm -f "$$log"' EXIT; for t in $$(find scripts/terraform/latitude-spx -maxdepth 1 -name '*_test.sh' | sort); do \
|
||||
$(_RUNSH); \
|
||||
done
|
||||
@$(_HDR) "\n....Running .github/scripts/*_test.sh...."
|
||||
@log=$$(mktemp); trap 'rm -f "$$log"' EXIT; for t in $$(find .github/scripts -maxdepth 1 -name '*_test.sh' | sort); do \
|
||||
$(_RUNSH); \
|
||||
done
|
||||
$(_Q)shellcheck -S warning .github/scripts/hypervisor-disk-preflight.sh \
|
||||
.github/scripts/wait-for-memory.sh \
|
||||
.github/scripts/go-build-cache.sh \
|
||||
.github/scripts/e2e-select.sh
|
||||
@$(_HDR) "\n....Running shellcheck over build/scripts/**/*.sh + the node scripts...."
|
||||
$(_Q)shellcheck -S warning $$(find build/scripts -name '*.sh' | sort) $(NODE_SCRIPTS)
|
||||
$(_Q)shellcheck -S warning -x scripts/build-oracle-appliance.sh $$(find scripts/appliance -name '*.sh' | sort)
|
||||
@$(_HDR) " test-build-scripts ok"
|
||||
@echo -e "\n....Running shellcheck over build/scripts/**/*.sh + the node scripts...."
|
||||
shellcheck -S warning $$(find build/scripts -name '*.sh' | sort) $(NODE_SCRIPTS)
|
||||
@echo " test-build-scripts ok"
|
||||
|
||||
# E2E harness unit tests. Build-tagged `e2e` so they're skipped by the
|
||||
# default `go test ./spinifex/...`. Runs with mocked AWS clients — no
|
||||
@@ -346,16 +224,6 @@ test-integration:
|
||||
@$(MAKE) --no-print-directory generate-aws-model-coverage
|
||||
@echo "AWS operation coverage: $(AWS_MODEL_OPERATION_COVERAGE_DIR)"
|
||||
|
||||
# Suites the unit tiers skip: the three-node SQS R3 scenarios behind
|
||||
# SPINIFEX_SQS_R3 and the lambdatestlink-tagged tests. CI runs these on every PR.
|
||||
test-gated:
|
||||
@echo -e "\n....Running env- and tag-gated suites...."
|
||||
$(_Q)LOG_IGNORE=1 SPINIFEX_SQS_R3=1 go test $(GOTESTFLAGS) -timeout 180s -run '^TestR3$$' ./spinifex/domains/sqs/message/jetstream/ $(_RACEQ)
|
||||
$(_Q)LOG_IGNORE=1 go test $(GOTESTFLAGS) -tags=lambdatestlink -timeout 120s ./cmd/lambda-agent/ ./spinifex/agents/lambda/hostlink/ ./spinifex/runtime/compute/lambda/guestconn/ ./spinifex/runtime/compute/lambda/testlink/ ./spinifex/domains/network/host/ $(_RACEQ)
|
||||
$(_Q)LOG_IGNORE=1 go test $(GOTESTFLAGS) -tags=lambdatestlink -timeout 120s -run '^TestLambdaTestLink(Executor|Reply)_' ./spinifex/runtime/roles/awsgw/ $(_RACEQ)
|
||||
$(_Q)LOG_IGNORE=1 go test $(GOTESTFLAGS) -tags=lambdatestlink,lambdatestlinkseam -timeout 120s -run '^TestLambdaTestLinkSeam_' ./spinifex/runtime/roles/awsgw/ $(_RACEQ)
|
||||
$(_Q)LOG_IGNORE=1 go test $(GOTESTFLAGS) -tags=lambdatestlink -timeout 120s -run '^TestLambdaTestLink' ./spinifex/runtime/daemon/ $(_RACEQ)
|
||||
|
||||
generate-aws-model-coverage:
|
||||
@go run ./cmd/aws-model-coverage -out $(AWS_MODEL_OPERATION_COVERAGE_DIR) -json $(AWS_MODEL_OPERATION_COVERAGE_DIR)/coverage.json
|
||||
|
||||
@@ -371,6 +239,23 @@ test-segscan-oracle:
|
||||
@echo -e "\n....Running segscan storage oracle test...."
|
||||
$(_Q)LOG_IGNORE=1 go test -tags=integration,segscanoracle -timeout 120s ./tests/integration/... -run TestSegscanOracle $(_RACEQ)
|
||||
|
||||
# Validate docs/service-interfaces.yaml. Schema check + cross-reference
|
||||
# of services/suites/fixtures + on-disk path existence.
|
||||
manifest-check:
|
||||
@echo -e "\n....Checking service-interfaces.yaml...."
|
||||
@go run ./tests/e2e/manifest-check/cmd/manifest-check -repo-root . -manifest docs/service-interfaces.yaml
|
||||
|
||||
# Drift guards: direct-create fixture lint + NATS subject lint,
|
||||
# ratcheted against tests/e2e/manifest-lint/baseline.txt. Fails only on NEW
|
||||
# drift beyond the baseline.
|
||||
manifest-lint:
|
||||
@echo -e "\n....Linting manifest drift (fixtures + subjects)...."
|
||||
@go run ./tests/e2e/manifest-lint/cmd/manifest-lint -repo-root .
|
||||
|
||||
# Accept current drift into the baseline. Run after an intentional change.
|
||||
manifest-lint-update:
|
||||
@go run ./tests/e2e/manifest-lint/cmd/manifest-lint -repo-root . -update
|
||||
|
||||
# Run unit tests
|
||||
test:
|
||||
@echo -e "\n....Running tests for $(GO_PROJECT_NAME)...."
|
||||
@@ -583,7 +468,6 @@ distro-amd64:
|
||||
echo "[distro-amd64] WARNING: microvm artifacts missing — tarball will not include them"; \
|
||||
echo "[distro-amd64] run 'make build-microvm-image' before 'make distro-amd64'"; \
|
||||
fi
|
||||
./scripts/lib/stage-sqs-store-microvm.sh $(SQS_WORKER_MICROVM_OUT_DIR) dist/amd64 sqs-worker
|
||||
tar -czf dist/spinifex-$(VERSION)-linux-amd64.tar.gz -C dist/amd64 .
|
||||
sha256sum dist/spinifex-$(VERSION)-linux-amd64.tar.gz > dist/spinifex-$(VERSION)-linux-amd64.tar.gz.sha256
|
||||
|
||||
@@ -602,7 +486,7 @@ distro-arm64:
|
||||
distro-clean:
|
||||
rm -rf dist/
|
||||
|
||||
.PHONY: test-package-check build build-lambda-testlink build-lambda-testlink-image build-ui test-ui-e2e ui-e2e-container ui-baselines build-installer build-lb-agent build-ecs-agent build-system-image build-eks-node-image import-eks-node-image publish-eks-node-image build-ecs-node-image import-ecs-node-image build-rds-postgres-image import-rds-postgres-image build-rds-mariadb-image import-rds-mariadb-image build-microvm-image build-sqs-worker-microvm-image build-lambda-runtime-microvm-image install-microvm install-sqs-worker-microvm go_build preflight test test-cover test-race diff-coverage bench fuzz test-actions test-images test-build-scripts test-harness test-integration test-gated generate-aws-model-coverage aws-model-coverage test-segscan-oracle \
|
||||
.PHONY: test-package-check build build-ui build-installer build-lb-agent build-ecs-agent build-system-image build-eks-node-image import-eks-node-image publish-eks-node-image build-ecs-node-image import-ecs-node-image build-rds-postgres-image import-rds-postgres-image build-rds-mariadb-image import-rds-mariadb-image build-microvm-image install-microvm go_build preflight test test-cover test-race diff-coverage bench fuzz test-actions test-images test-build-scripts test-harness test-integration generate-aws-model-coverage aws-model-coverage test-segscan-oracle manifest-check manifest-lint manifest-lint-update \
|
||||
deploy reinstall clean \
|
||||
install-system install-go install-aws quickinstall \
|
||||
lint fix govulncheck nilaway \
|
||||
|
||||
@@ -54,7 +54,7 @@ For Terraform, start from the [example workbooks](https://docs.mulgadc.com/docs/
|
||||
|
||||
## Services
|
||||
|
||||
The platform serves **405 AWS API operations** across the services below. See exactly which ones in the [coverage report](https://docs.mulgadc.com/coverage).
|
||||
The platform serves **392 AWS API operations** across the services below. See exactly which ones in the [coverage report](https://docs.mulgadc.com/coverage).
|
||||
|
||||
| Service | What it is | Built on | Status |
|
||||
| --- | --- | --- | --- |
|
||||
@@ -68,10 +68,8 @@ The platform serves **405 AWS API operations** across the services below. See ex
|
||||
| **ECR** | Container registry | Predastore, deduplicated per account | Available |
|
||||
| **ECS** | Container service | Spinifex instances | Available |
|
||||
| **RDS** | Databases | PostgreSQL and MariaDB, with automated backups | Available |
|
||||
| **Route 53** | DNS | [Northstar](https://github.com/mulgadc/northstar): authoritative DNS with hosted zones | Available |
|
||||
| **Bedrock** | AI deployment | vLLM for text generation and Hugging Face TEI for embeddings and reranking | Q4 2026 |
|
||||
| **Lambda** | Serverless functions | QEMU `microvm` execution environments | Q1 2027 |
|
||||
| **SQS** | Message queues | A dedicated NATS JetStream store in a per-Region QEMU `microvm` guest | Q1 2027 |
|
||||
| **Lambda** | Serverless functions | | Q1 2027 |
|
||||
|
||||
Roadmap services ship under the same AWS API surface, so code written for AWS today works the moment they land. [Track what's shipped in the release notes](https://github.com/mulgadc/spinifex/releases).
|
||||
|
||||
@@ -113,7 +111,7 @@ aws ec2 describe-instance-types
|
||||
## Web Console
|
||||
|
||||
<p align="center">
|
||||
<img src=".github/assets/dashboard.webp" alt="Spinifex web console dashboard" width="900">
|
||||
<img src=".github/assets/spinifex-ui.jpg" alt="Spinifex web console dashboard" width="900">
|
||||
</p>
|
||||
|
||||
Spinifex ships with a built-in web console for your instances, volumes, buckets, networks, and IAM. See [Launching the Web UI](https://docs.mulgadc.com/docs/setting-up-your-cluster#7-launching-the-web-ui).
|
||||
|
||||
@@ -15,8 +15,12 @@ var allowlist = []allowance{
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/igw", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/image", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/instance", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/key", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/launchtemplate", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/placementgroup", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/routetable", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/snapshot", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/volume", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
{From: "operator/accountlifecycle", To: "domains/ec2/awsapi/vpc", Debt: "accountteardown (dependency)", Reason: "EC2 reapers call the awsapi action adapters; needs domain-owned cleanup capabilities"},
|
||||
|
||||
{From: "agents/ecs/agent", To: "domains/ecs", Debt: "agents/ecs/agent", Reason: "assignment/state wire types from the ECS implementation; needs a guest/controller contract"},
|
||||
@@ -25,17 +29,8 @@ var allowlist = []allowance{
|
||||
|
||||
{From: "agents/rds/agent", To: "domains/rds/instance", Debt: "agents/rds/agent (dependency, narrowed)", Reason: "command protocol, wire and record types; needs a guest/controller wire contract"},
|
||||
|
||||
{From: "cmd/aws-model-coverage", To: "domains/acm/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/ecr/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/route53/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/elbv2/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/rds/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/ec2/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/ecs/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/eks/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/iam/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/sqs/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/sts/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared service inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/acm/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared ECR/ACM inventories; named exception, must not construct a registration"},
|
||||
{From: "cmd/aws-model-coverage", To: "domains/ecr/awsapi", Debt: "cmd/aws-model-coverage", Reason: "reads declared ECR/ACM inventories; named exception, must not construct a registration"},
|
||||
|
||||
{From: "runtime/daemon", To: "domains/ec2/awsapi/instance", Debt: "Instance adapter consumers outside EC2", Reason: "runtime/daemon/eks_cp_control.go and rds_deps.go call the instance adapter; needs an EC2 capability"},
|
||||
|
||||
@@ -57,6 +52,9 @@ var allowlist = []allowance{
|
||||
{From: "domains/ec2/awsapi/instance", To: "runtime/compute/cache", Debt: "domains/ec2/awsapi/instance (dependency)", Reason: "node liveness and VM state; needs an EC2-owned projection"},
|
||||
{From: "domains/ec2/awsapi/instance", To: "runtime/compute/vm", Debt: "domains/ec2/awsapi/instance (dependency)", Reason: "node liveness and VM state; needs an EC2-owned projection"},
|
||||
|
||||
{From: "domains/ec2/awsapi/spotinstance", To: "domains/admission/quota", Debt: "domains/ec2/awsapi/spotinstance (dependency)", Reason: "RequestSpotInstances.go uses IAMService and quota.Service"},
|
||||
{From: "domains/ec2/awsapi/spotinstance", To: "domains/iam", Debt: "domains/ec2/awsapi/spotinstance (dependency)", Reason: "RequestSpotInstances.go uses IAMService and quota.Service"},
|
||||
|
||||
{From: "domains/ec2/eip", To: "domains/network/topology", Debt: "EC2 → network implementation", Reason: "EC2 reads network implementation; clears with the EC2-owned projection"},
|
||||
|
||||
{From: "domains/ec2/guestmetadata", To: "domains/dns", Debt: "Guest metadata: the cross-domain and runtime imports above", Reason: "guest-metadata relocation residual; needs capabilities, projections or contracts"},
|
||||
@@ -93,14 +91,42 @@ var allowlist = []allowance{
|
||||
{From: "domains/ochre", To: "domains/ec2/systeminstance", Debt: "Cross-domain", Reason: "consumes another domain's implementation rather than a capability"},
|
||||
{From: "domains/ochre", To: "domains/iam", Debt: "IAM implementation consumers in STS, Ochre and ELBv2", Reason: "secret encrypt/decrypt helpers; needs a shared crypto boundary"},
|
||||
{From: "domains/ochre", To: "domains/network/systemvpc", Debt: "Cross-domain", Reason: "consumes another domain's implementation rather than a capability"},
|
||||
{From: "domains/ochre", To: "ingress/aws/bodyscope", Debt: "domains/ochre (action adapters)", Reason: "Bedrock action adapters still live in the Ochre package; they move to domains/ochre/awsapi"},
|
||||
{From: "domains/ochre", To: "runtime/compute/gpu", Debt: "Domain → runtime", Reason: "GPU runtime state read directly; needs a domain-owned projection"},
|
||||
|
||||
{From: "domains/ochre/vector", To: "domains/iam", Debt: "Cross-domain", Reason: "consumes another domain's implementation rather than a capability"},
|
||||
|
||||
{From: "gateway", To: "domains/ec2/awsapi", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/account", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/capacityreservation", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/eigw", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/eip", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/idem", Debt: "gateway/ec2.go (registration; ec2_idempotency.go)", Reason: "dispatch-time idempotency wrapper; moves with the EC2 dispatch map"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/igw", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/image", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/instance", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/key", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/launchtemplate", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/natgw", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/placementgroup", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/routetable", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/snapshot", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/spotinstance", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/tags", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/volume", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/vpc", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ec2/awsapi/zone", Debt: "gateway/ec2.go (registration)", Reason: "EC2 dispatch map stays in gateway until the ingress registration seam"},
|
||||
|
||||
{From: "domains/rds/awsapi", To: "domains/ec2/awsapi/instance", Debt: "Instance adapter consumers outside EC2", Reason: "enginecatalog.go calls the instance adapter; needs an EC2 capability"},
|
||||
|
||||
{From: "domains/sts/awsapi", To: "domains/iam", Debt: "IAM implementation consumers in STS, Ochre and ELBv2", Reason: "GetCallerIdentity takes IAMService; needs an IAM capability"},
|
||||
|
||||
{From: "gateway", To: "domains/eks/awsapi", Debt: "gateway (service registration)", Reason: "IAM, STS, EKS and RDS dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/iam/awsapi", Debt: "gateway (service registration)", Reason: "IAM, STS, EKS and RDS dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/rds/awsapi", Debt: "gateway (service registration)", Reason: "IAM, STS, EKS and RDS dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/sts/awsapi", Debt: "gateway (service registration)", Reason: "IAM, STS, EKS and RDS dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/ecs/awsapi", Debt: "gateway (service registration)", Reason: "ECS and ELBv2 dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "gateway", To: "domains/elbv2/awsapi", Debt: "gateway (service registration)", Reason: "ECS and ELBv2 dispatch stays in gateway until the ingress registration seam"},
|
||||
{From: "domains/ecs", To: "domains/ec2/ebs/metadata", Debt: "close-out: handlers/ecs (ADR-0004 blocked-dependency list; inventory-ecs boundary imports)", Reason: "legacy ECS calls another domain's implementation as an internal API"},
|
||||
{From: "domains/ecs", To: "domains/ec2/eip", Debt: "close-out: handlers/ecs (ADR-0004 blocked-dependency list; inventory-ecs boundary imports)", Reason: "legacy ECS calls another domain's implementation as an internal API"},
|
||||
{From: "domains/ecs", To: "domains/ec2/instancetypes", Debt: "close-out: handlers/ecs (ADR-0004 blocked-dependency list; inventory-ecs boundary imports)", Reason: "legacy ECS calls another domain's implementation as an internal API"},
|
||||
@@ -133,4 +159,6 @@ var allowlist = []allowance{
|
||||
{From: "domains/rds/instance", To: "runtime/compute/vm", Debt: "close-out: handlers/rds (ADR-0004 blocked-dependency list; inventory-rds boundary imports)", Reason: "legacy RDS calls another domain's or runtime implementation directly"},
|
||||
|
||||
{From: "domains/sts", To: "domains/iam", Debt: "IAM implementation consumers in STS, Ochre and ELBv2", Reason: "STS reads IAM service, trust-policy and OIDC-provider state; needs an IAM capability"},
|
||||
|
||||
{From: "operator/cli", To: "gateway", Debt: "operator/cli (dependency)", Reason: "command tree imports implementations; needs public capabilities"},
|
||||
}
|
||||
|
||||
@@ -192,47 +192,24 @@ func TestADR0001_LayeringCheckerDetectsFaults(t *testing.T) {
|
||||
}
|
||||
}
|
||||
|
||||
// TestADR0001_R2_IngressRejectsRuntimeOperatorAndDomains: the request pipeline
|
||||
// lives under ingress/aws, so it must not import the daemon, any runtime or
|
||||
// operator package, or a domain implementation.
|
||||
func TestADR0001_R2_IngressRejectsRuntimeOperatorAndDomains(t *testing.T) {
|
||||
r := ruleByID("R2_IngressImportsNoImplementation")
|
||||
from := "spinifex/ingress/aws/pipeline"
|
||||
f, ok := classify(from)
|
||||
if !ok || !r.subject(f, from) {
|
||||
t.Fatalf("%s is not governed by R2", from)
|
||||
// TestADR0001_InterimDaemonDirectionCoversRuntimeDaemon: the daemon left its
|
||||
// legacy root, so gateway importing runtime/daemon must still be rejected, and
|
||||
// the former handlers packages are held to it by R5 at their domain homes.
|
||||
func TestADR0001_InterimDaemonDirectionCoversRuntimeDaemon(t *testing.T) {
|
||||
daemon, _ := classify("spinifex/runtime/daemon")
|
||||
checks := []struct{ rule, from string }{
|
||||
{"Interim_DaemonGatewayHandlersDirection", "spinifex/gateway"},
|
||||
{"R5_DomainImportsNoRuntimeOrOperator", "spinifex/domains/ecs"},
|
||||
}
|
||||
for _, to := range []string{
|
||||
"spinifex/runtime/daemon", "spinifex/runtime/host/process", "spinifex/runtime/roles/awsgw",
|
||||
"spinifex/operator/admin", "spinifex/domains/ecs", "spinifex/domains/iam/awsapi",
|
||||
} {
|
||||
target, _ := classify(to)
|
||||
if !r.forbids(f, target, from) {
|
||||
t.Errorf("%s -> %s is not rejected by R2", from, to)
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
// TestADR0002_S7_RequestRecordRejectsOutsiders: the request record is readable
|
||||
// by packages under ingress/aws only.
|
||||
func TestADR0002_S7_RequestRecordRejectsOutsiders(t *testing.T) {
|
||||
record, ok := classify("spinifex/ingress/aws/reqctx")
|
||||
if !ok || !record.requestRecord {
|
||||
t.Fatalf("ingress/aws/reqctx is not classified as the request record: %+v", record)
|
||||
}
|
||||
r := ruleByID("S7_RequestRecordPrivateToIngress")
|
||||
for name, wantRejected := range map[string]bool{
|
||||
"domains/ecs/awsapi": true,
|
||||
"runtime/roles/awsgw": true,
|
||||
"operator/admin": true,
|
||||
"ingress/aws/dispatch": false,
|
||||
} {
|
||||
from, ok := classify(fullPath(name))
|
||||
if !ok {
|
||||
t.Fatalf("%s is unclassified", name)
|
||||
}
|
||||
if got := r.forbids(from, record, name); got != wantRejected {
|
||||
t.Errorf("%s -> ingress/aws/reqctx rejected=%v, want %v", name, got, wantRejected)
|
||||
for _, c := range checks {
|
||||
r := ruleByID(c.rule)
|
||||
f, _ := classify(c.from)
|
||||
if !r.subject(f, c.from) || !r.forbids(f, daemon, c.from) {
|
||||
t.Errorf("%s -> runtime/daemon is not rejected by %s", c.from, c.rule)
|
||||
}
|
||||
}
|
||||
other, _ := classify("spinifex/runtime/host/process")
|
||||
if gw, _ := classify("spinifex/gateway"); ruleByID("Interim_DaemonGatewayHandlersDirection").forbids(gw, other, "spinifex/gateway") {
|
||||
t.Error("gateway -> runtime/host/process is rejected, but it is not the daemon")
|
||||
}
|
||||
}
|
||||
|
||||
@@ -29,11 +29,12 @@ const (
|
||||
type domainKind int
|
||||
|
||||
const (
|
||||
notDomain domainKind = iota
|
||||
domainImpl // domains/<d>/... resource implementation
|
||||
domainAWSAPI // domains/<d>/awsapi/... AWS action adapters
|
||||
domainProj // domains/<d>/projection/... authorised view
|
||||
legacyImpl // handlers/<d>/... legacy domain implementation
|
||||
notDomain domainKind = iota
|
||||
domainImpl // domains/<d>/... resource implementation
|
||||
domainAWSAPI // domains/<d>/awsapi/... AWS action adapters
|
||||
domainProj // domains/<d>/projection/... authorised view
|
||||
legacyImpl // handlers/<d>/... legacy domain implementation
|
||||
legacyAdapter // gateway/<d>/... legacy AWS action adapters
|
||||
)
|
||||
|
||||
type class struct {
|
||||
@@ -42,8 +43,6 @@ type class struct {
|
||||
domain string // owning domain for domain-shaped and agent packages
|
||||
kind domainKind
|
||||
daemon bool // runtime/daemon, the interim daemon -> gateway -> handlers root
|
||||
// requestRecord marks ingress/aws/reqctx, the request record ADR-0002:S7 keeps private to ingress.
|
||||
requestRecord bool
|
||||
}
|
||||
|
||||
func (c class) domainShaped() bool { return c.kind != notDomain }
|
||||
@@ -59,8 +58,17 @@ func (c class) isLegacy(roots ...string) bool {
|
||||
return slices.Contains(roots, c.legacy)
|
||||
}
|
||||
|
||||
// legacyRoots are pre-ADR-0001 application roots still standing. None remain.
|
||||
var legacyRoots = map[string]bool{}
|
||||
// legacyRoots are pre-ADR-0001 application roots still standing.
|
||||
var legacyRoots = map[string]bool{
|
||||
"gateway": true,
|
||||
}
|
||||
|
||||
// gatewayDomains names the domain each gateway/<service> adapter belongs to;
|
||||
// Bedrock is the AWS surface of the Ochre domain.
|
||||
var gatewayDomains = map[string]string{
|
||||
"bedrock": "ochre", "ecs": "ecs", "eks": "eks", "elbv2": "elbv2", "iam": "iam",
|
||||
"rds": "rds", "sts": "sts",
|
||||
}
|
||||
|
||||
// classify maps a module-relative package directory to its layer. ok is false
|
||||
// for a package that has no deliberate classification.
|
||||
@@ -95,7 +103,7 @@ func classifyApp(app []string) (class, bool) {
|
||||
case "foundation":
|
||||
return class{layer: layerFoundation}, true
|
||||
case "ingress":
|
||||
return class{layer: layerIngress, requestRecord: len(app) >= 3 && app[1] == "aws" && app[2] == "reqctx"}, true
|
||||
return class{layer: layerIngress}, true
|
||||
case "runtime":
|
||||
return class{layer: layerRuntime, daemon: len(app) >= 2 && app[1] == "daemon"}, true
|
||||
case "operator":
|
||||
@@ -132,6 +140,8 @@ func classifyApp(app []string) (class, bool) {
|
||||
switch {
|
||||
case app[0] == "handlers" && len(app) >= 2:
|
||||
c.domain, c.kind = app[1], legacyImpl
|
||||
case app[0] == "gateway" && len(app) >= 2 && gatewayDomains[app[1]] != "":
|
||||
c.domain, c.kind = gatewayDomains[app[1]], legacyAdapter
|
||||
}
|
||||
return c, true
|
||||
}
|
||||
|
||||
@@ -9,11 +9,14 @@ import (
|
||||
// frozenLegacyPackages is every production package still under a legacy root.
|
||||
// It only shrinks: a new entry is a new legacy package, a missing one is a
|
||||
// completed move whose line must be deleted.
|
||||
var frozenLegacyPackages = []string{}
|
||||
var frozenLegacyPackages = []string{
|
||||
"gateway",
|
||||
"gateway/spx",
|
||||
}
|
||||
|
||||
// retiredLegacyRoots were legacy roots whose every package has moved to its
|
||||
// ADR-0001 target; recreating one would reopen a closed migration row.
|
||||
var retiredLegacyRoots = []string{"accountteardown", "admin", "daemon", "gateway", "handlers", "lbagent", "services", "utils", "vpcd"}
|
||||
var retiredLegacyRoots = []string{"accountteardown", "admin", "daemon", "handlers", "lbagent", "services", "utils", "vpcd"}
|
||||
|
||||
// TestADR0001_LegacyRootsFrozen: the current-to-target map gives every legacy
|
||||
// root a destination, so none may gain a package and no retired root may return.
|
||||
@@ -33,6 +36,9 @@ func TestADR0001_LegacyRootsFrozen(t *testing.T) {
|
||||
present[name] = true
|
||||
}
|
||||
}
|
||||
if len(present) == 0 {
|
||||
t.Fatal("found no legacy packages; the freeze would pass vacuously")
|
||||
}
|
||||
frozen := map[string]bool{}
|
||||
for _, p := range frozenLegacyPackages {
|
||||
frozen[p] = true
|
||||
|
||||
@@ -1,9 +1,6 @@
|
||||
package layering
|
||||
|
||||
import (
|
||||
"slices"
|
||||
"strings"
|
||||
)
|
||||
import "slices"
|
||||
|
||||
// rule is one ADR-0001 dependency rule over a classified import edge.
|
||||
type rule struct {
|
||||
@@ -126,8 +123,10 @@ var rules = []rule{
|
||||
case t.layer == layerAgents:
|
||||
return true
|
||||
case t.layer == layerIngress:
|
||||
return f.kind != domainAWSAPI
|
||||
case t.isLegacy("handlers"):
|
||||
return f.kind != domainAWSAPI && f.kind != legacyAdapter
|
||||
case t.isLegacy("gateway") && !t.domainShaped():
|
||||
return true
|
||||
case t.isLegacy("handlers", "gateway"):
|
||||
return f.layer == layerDomain
|
||||
}
|
||||
return false
|
||||
@@ -153,15 +152,18 @@ var rules = []rule{
|
||||
},
|
||||
},
|
||||
{
|
||||
id: "S7_RequestRecordPrivateToIngress",
|
||||
clause: "ADR-0002:S7: the request record is private to ingress, so no domain, runtime or operator package reads it; only packages under ingress/aws may import ingress/aws/reqctx",
|
||||
hint: "receive what you need through dispatch.Invocation instead of the request record",
|
||||
subject: func(_ class, _ string) bool { return true },
|
||||
forbids: func(_, t class, fromName string) bool {
|
||||
if !t.requestRecord {
|
||||
return false
|
||||
id: "Interim_DaemonGatewayHandlersDirection",
|
||||
clause: "during migration the existing daemon -> gateway -> handlers rule remains the guardrail (Q-20, Q-29)",
|
||||
hint: "keep the dependency pointing daemon -> gateway -> handlers; invert it through a callback or capability",
|
||||
subject: func(f class, _ string) bool { return f.isLegacy("handlers", "gateway") },
|
||||
forbids: func(f, t class, _ string) bool {
|
||||
switch {
|
||||
case f.isLegacy("handlers"):
|
||||
return t.isLegacy("gateway") || t.daemon
|
||||
case f.isLegacy("gateway"):
|
||||
return t.daemon
|
||||
}
|
||||
return !strings.HasPrefix(fromName, "ingress/aws/")
|
||||
return false
|
||||
},
|
||||
},
|
||||
}
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
ARG VERSION=dev
|
||||
|
||||
FROM golang:1.27.3-bookworm@sha256:d9b6a1beb9474adc2a1ce842573639b44b2a0dd49b6669f51d3821baf9ec37c9 AS builder
|
||||
FROM golang:1.27.3-bookworm@sha256:972b4110ba0f47539a975c31e52c64765e1ff19f2d74469ae4a595c6676e80a5 AS builder
|
||||
|
||||
ARG VERSION
|
||||
|
||||
|
||||
@@ -32,8 +32,6 @@ load_mod qemu_fw_cfg || load_mod fw_cfg_sysfs || true
|
||||
load_mod failover || true
|
||||
load_mod net_failover || true
|
||||
load_mod virtio_net || true
|
||||
# lb-agent probes each data NIC's gateway on an AF_PACKET socket.
|
||||
load_mod af_packet || true
|
||||
|
||||
NETCFG=/sys/firmware/qemu_fw_cfg/by_name/opt/spinifex/netcfg/raw
|
||||
i=0
|
||||
|
||||
@@ -1,90 +0,0 @@
|
||||
#!/bin/sh
|
||||
mount -t proc proc /proc
|
||||
mount -t sysfs sys /sys
|
||||
mount -t devtmpfs dev /dev
|
||||
|
||||
# modprobe needs a usable modules.dep; insmod by explicit path does not, so
|
||||
# fall back to it. Handles .ko and .ko.gz.
|
||||
load_mod() {
|
||||
modprobe "$1" 2>/dev/null && return 0
|
||||
for ko in $(find /lib/modules -name "$1.ko" -o -name "$1.ko.gz" 2>/dev/null); do
|
||||
insmod "$ko" 2>/dev/null && return 0
|
||||
case "$ko" in
|
||||
*.gz) gunzip -c "$ko" >/tmp/mod.ko 2>/dev/null \
|
||||
&& insmod /tmp/mod.ko 2>/dev/null && return 0 ;;
|
||||
esac
|
||||
done
|
||||
return 1
|
||||
}
|
||||
# lambda-agent reads its identity from fw_cfg, or the configuration disk, and
|
||||
# seals it before anything else runs. virtio_net needs net_failover and failover, and the
|
||||
# insmod fallback resolves no dependencies, so load the chain bottom-up.
|
||||
load_mod qemu_fw_cfg || echo "[init] qemu_fw_cfg did not load" >/dev/console
|
||||
load_mod failover || true
|
||||
load_mod net_failover || true
|
||||
load_mod virtio_net || echo "[init] virtio_net did not load" >/dev/console
|
||||
|
||||
ip link set lo up
|
||||
|
||||
# Without fw_cfg data the executor attaches the same entries as a tar archive
|
||||
# on the second disk; lambda-agent reads them here and erases the disk before
|
||||
# the function starts. qemu_fw_cfg loads without a device, so test the entry.
|
||||
CFG_ROOT=/spinifex-cfg
|
||||
FWCFG_ROOT=/sys/firmware/qemu_fw_cfg
|
||||
if [ ! -f "$FWCFG_ROOT/by_name/opt/spinifex/netcfg/raw" ]; then
|
||||
load_mod virtio_blk || echo "[init] virtio_blk did not load" >/dev/console
|
||||
i=0
|
||||
while [ "$i" -lt 25 ] && [ ! -b /dev/vdb ]; do
|
||||
sleep 0.04
|
||||
i=$((i + 1))
|
||||
done
|
||||
if [ -b /dev/vdb ]; then
|
||||
mkdir -m 0700 "$CFG_ROOT"
|
||||
if ! mount -t tmpfs -o mode=0700,size=1m tmpfs "$CFG_ROOT" ||
|
||||
! tar -xf /dev/vdb -C "$CFG_ROOT"; then
|
||||
echo "[init] configuration disk not readable" >/dev/console
|
||||
fi
|
||||
FWCFG_ROOT=$CFG_ROOT
|
||||
fi
|
||||
fi
|
||||
|
||||
# The guest's addresses come from the same netcfg entry as other system VMs.
|
||||
NETCFG=$FWCFG_ROOT/by_name/opt/spinifex/netcfg/raw
|
||||
i=0
|
||||
while [ "$i" -lt 25 ] && [ ! -f "$NETCFG" ]; do
|
||||
sleep 0.04
|
||||
i=$((i + 1))
|
||||
done
|
||||
if [ -f "$NETCFG" ]; then
|
||||
# shellcheck source=/dev/null
|
||||
. "$NETCFG"
|
||||
for n in 0 1 2 3 4 5; do
|
||||
eval mac=\$NIC${n}_MAC
|
||||
[ -z "$mac" ] && continue
|
||||
iface=""
|
||||
for _ in 1 2 3 4 5; do
|
||||
iface=$(ip -o link | awk -v m="$mac" 'tolower($0) ~ tolower(m) {print $2}' | tr -d :)
|
||||
[ -n "$iface" ] && break
|
||||
sleep 0.02
|
||||
done
|
||||
if [ -z "$iface" ]; then
|
||||
echo "[init] no interface for MAC $mac (NIC$n)" >/dev/console
|
||||
continue
|
||||
fi
|
||||
cidr="" gw="" default="" rdst="" rvia=""
|
||||
eval cidr=\$NIC${n}_CIDR
|
||||
eval gw=\$NIC${n}_GW
|
||||
eval default=\$NIC${n}_DEFAULT
|
||||
eval rdst=\$NIC${n}_ROUTE_DST
|
||||
eval rvia=\$NIC${n}_ROUTE_VIA
|
||||
ip link set "$iface" up
|
||||
[ -n "$cidr" ] && ip addr add "$cidr" dev "$iface"
|
||||
[ "$default" = "1" ] && [ -n "$gw" ] && ip route add default via "$gw" dev "$iface"
|
||||
[ -n "$rdst" ] && [ -n "$rvia" ] && ip route add "$rdst" via "$rvia" dev "$iface"
|
||||
done
|
||||
else
|
||||
# lambda-agent then fails to reach the host and the guest powers off.
|
||||
echo "[init] fw_cfg netcfg not available" >/dev/console
|
||||
fi
|
||||
|
||||
exec /sbin/init
|
||||
@@ -1,6 +0,0 @@
|
||||
# The power button loads here, not in /init: busybox init has its SIGUSR2
|
||||
# (poweroff) handler installed before sysinit runs, so a press can never reach
|
||||
# the pre-exec shell PID 1. A press before this line is lost; the host re-sends.
|
||||
::sysinit:/sbin/modprobe tiny_power_button power_signal=12
|
||||
::once:/usr/local/lib/lambda-runtime/run-agent
|
||||
::ctrlaltdel:/sbin/reboot
|
||||
@@ -1,6 +0,0 @@
|
||||
#!/bin/sh
|
||||
# Runs the Lambda guest agent, then powers the guest off whatever its status.
|
||||
/usr/local/bin/lambda-agent >/dev/console 2>&1
|
||||
status=$?
|
||||
echo "[run-agent] lambda-agent exited with status $status" >/dev/console
|
||||
poweroff -f
|
||||
@@ -1,100 +0,0 @@
|
||||
#!/bin/sh
|
||||
mount -t proc proc /proc
|
||||
mount -t sysfs sys /sys
|
||||
mount -t devtmpfs dev /dev
|
||||
|
||||
# virtio_mmio is built-in (CONFIG_VIRTIO_MMIO=y) so virtio-net devices are
|
||||
# already enumerated by the time PID 1 runs — no modprobe needed.
|
||||
# qemu_fw_cfg and virtio_net are =m in stock Alpine linux-virt, so load them.
|
||||
#
|
||||
# modprobe needs a usable modules.dep; when the image is built on a host whose
|
||||
# depmod can't produce one, busybox modprobe silently no-ops and the module
|
||||
# never loads — fw_cfg then never appears and init panics. insmod by explicit
|
||||
# path needs no modules.dep, so fall back to it. Handles .ko and .ko.gz.
|
||||
load_mod() {
|
||||
modprobe "$1" 2>/dev/null && return 0
|
||||
for ko in $(find /lib/modules -name "$1.ko" -o -name "$1.ko.gz" 2>/dev/null); do
|
||||
insmod "$ko" 2>/dev/null && return 0
|
||||
case "$ko" in
|
||||
*.gz) gunzip -c "$ko" >/tmp/mod.ko 2>/dev/null \
|
||||
&& insmod /tmp/mod.ko 2>/dev/null && return 0 ;;
|
||||
esac
|
||||
done
|
||||
return 1
|
||||
}
|
||||
load_mod qemu_fw_cfg || load_mod fw_cfg_sysfs || true
|
||||
# virtio_net depends on net_failover, which depends on failover. The load_mod
|
||||
# insmod fallback loads a single .ko with no dependency resolution, and
|
||||
# modules.dep is absent when the image is built on a kmod-less host (depmod
|
||||
# skipped), so modprobe can't resolve the chain either. Load it explicitly
|
||||
# bottom-up so the NIC comes up regardless of how the image was built —
|
||||
# without this the agent has no network and the LB hangs in provisioning.
|
||||
load_mod failover || true
|
||||
load_mod net_failover || true
|
||||
load_mod virtio_net || true
|
||||
|
||||
# The guest's SQS store listens on 127.0.0.1 and the worker dials it there;
|
||||
# the kernel leaves lo down, so bring it up before init starts either.
|
||||
ip link set lo up
|
||||
|
||||
NETCFG=/sys/firmware/qemu_fw_cfg/by_name/opt/spinifex/netcfg/raw
|
||||
i=0
|
||||
while [ "$i" -lt 25 ]; do
|
||||
[ -f "$NETCFG" ] && break
|
||||
sleep 0.04
|
||||
i=$((i + 1))
|
||||
done
|
||||
|
||||
if [ ! -f "$NETCFG" ]; then
|
||||
# Dump host/guest fw_cfg state to the serial console before init exits so a
|
||||
# "netcfg not available" panic is diagnosable from the captured console log
|
||||
# (module-not-loaded vs qemu-not-exposing-by_name vs wrong machine type).
|
||||
echo "[init] FATAL: fw_cfg netcfg not available" >&2
|
||||
echo "[init] dmi: vendor=$(cat /sys/class/dmi/id/sys_vendor 2>/dev/null) product=$(cat /sys/class/dmi/id/product_name 2>/dev/null) bios=$(cat /sys/class/dmi/id/bios_version 2>/dev/null)" >&2
|
||||
echo "[init] qemu_fw_cfg dir present: $([ -d /sys/firmware/qemu_fw_cfg ] && echo yes || echo NO)" >&2
|
||||
echo "[init] modules: $(lsmod 2>/dev/null | awk 'NR>1{print $1}' | tr '\n' ',')" >&2
|
||||
echo "[init] by_name listing:" >&2
|
||||
ls -la /sys/firmware/qemu_fw_cfg/by_name/ 2>&1 | sed 's/^/[init] /' >&2
|
||||
ls -la /sys/firmware/qemu_fw_cfg/by_name/opt/ 2>&1 | sed 's/^/[init] /' >&2
|
||||
echo "[init] dmesg fw_cfg:" >&2
|
||||
dmesg 2>/dev/null | grep -i fw_cfg | sed 's/^/[init] /' >&2
|
||||
exit 1
|
||||
fi
|
||||
|
||||
. "$NETCFG"
|
||||
|
||||
for n in 0 1 2 3 4 5; do
|
||||
eval mac=\$NIC${n}_MAC
|
||||
[ -z "$mac" ] && continue
|
||||
|
||||
iface=""
|
||||
for attempt in 1 2 3 4 5; do
|
||||
iface=$(ip -o link | awk -v m="$mac" 'tolower($0) ~ tolower(m) {print $2}' | tr -d :)
|
||||
[ -n "$iface" ] && break
|
||||
sleep 0.02
|
||||
done
|
||||
|
||||
if [ -z "$iface" ]; then
|
||||
echo "[init] WARNING: no interface found for MAC $mac (NIC$n), skipping" >&2
|
||||
continue
|
||||
fi
|
||||
|
||||
eval cidr=\$NIC${n}_CIDR
|
||||
eval gw=\$NIC${n}_GW
|
||||
eval default=\$NIC${n}_DEFAULT
|
||||
eval rdst=\$NIC${n}_ROUTE_DST
|
||||
eval rvia=\$NIC${n}_ROUTE_VIA
|
||||
|
||||
ip link set "$iface" up
|
||||
[ -n "$cidr" ] && ip addr add "$cidr" dev "$iface"
|
||||
[ "$default" = "1" ] && [ -n "$gw" ] && ip route add default via "$gw" dev "$iface"
|
||||
[ -n "$rdst" ] && [ -n "$rvia" ] && ip route add "$rdst" via "$rvia" dev "$iface"
|
||||
done
|
||||
|
||||
# The image takes no lb-agent environment. The worker reads the cluster CA
|
||||
# from fw_cfg itself; this copy only serves other tools in the guest.
|
||||
CA_CERT=/sys/firmware/qemu_fw_cfg/by_name/opt/spinifex/ca-cert/raw
|
||||
mkdir -p /etc/ssl/certs
|
||||
[ -f "$CA_CERT" ] && cp "$CA_CERT" /etc/ssl/certs/ca-certificates.crt
|
||||
|
||||
exec /sbin/init
|
||||
@@ -1,9 +0,0 @@
|
||||
# The power button loads here, not in /init: busybox init has its SIGUSR2
|
||||
# (poweroff) handler installed before sysinit runs, so a press can never reach
|
||||
# the pre-exec shell PID 1. A press before this line is lost; the host re-sends.
|
||||
::sysinit:/sbin/modprobe tiny_power_button power_signal=12
|
||||
::sysinit:/usr/local/lib/sqs-worker/sqs-worker-prepare
|
||||
::respawn:/usr/local/lib/sqs-worker/sqs-worker-store
|
||||
::respawn:/usr/local/lib/sqs-worker/sqs-worker-ready
|
||||
::ctrlaltdel:/sbin/reboot
|
||||
::shutdown:/usr/local/lib/sqs-worker/sqs-worker-stop
|
||||
@@ -1,203 +0,0 @@
|
||||
# sqs-worker-lib.sh — shared by the SQS guest's store, readiness and stop
|
||||
# steps. Sourced, never executed. Every path and command is overridable so the
|
||||
# host test can drive the same code against test-owned processes and stub tools.
|
||||
#
|
||||
# The guest runs its own SQS nats-server on loopback and the worker beside it.
|
||||
# The server starts only after boot settled the receipt key; store-ready exists
|
||||
# only while the running watcher's last check of that server passed.
|
||||
|
||||
SQSW_RUN=${SQSW_RUN:-/run/sqs-worker}
|
||||
SQSW_DATA=${SQSW_DATA:-/data}
|
||||
SQSW_CONSOLE=${SQSW_CONSOLE:-/dev/console}
|
||||
SQSW_MOUNTS=${SQSW_MOUNTS:-/proc/mounts}
|
||||
SQSW_BIN=${SQSW_BIN:-/usr/local/bin/sqs-worker}
|
||||
SQSW_LIB=${SQSW_LIB:-/usr/local/lib/sqs-worker/sqs-worker-lib.sh}
|
||||
SQSW_NATS_BIN=${SQSW_NATS_BIN:-/usr/local/bin/nats-server}
|
||||
SQSW_STORE_PORT=${SQSW_STORE_PORT:-4222}
|
||||
SQSW_STOP_BUDGET_TENTHS=${SQSW_STOP_BUDGET_TENTHS:-100}
|
||||
SQSW_RESPAWN_GAP=${SQSW_RESPAWN_GAP:-5}
|
||||
SQSW_PARK=${SQSW_PARK:-sleep 2147483647}
|
||||
SQSW_BOOT_RETRY=${SQSW_BOOT_RETRY:-30}
|
||||
UMOUNT=${UMOUNT:-umount}
|
||||
|
||||
sqsw_log() { echo "[sqs-worker] $*" >> "$SQSW_CONSOLE"; }
|
||||
|
||||
# sqsw_alive <pid>: running or stopped, not exited (a zombie counts as exited).
|
||||
sqsw_alive() { [ -r "/proc/$1/status" ] && ! grep -q '^State:[[:space:]]*Z' "/proc/$1/status" 2>/dev/null; }
|
||||
|
||||
# sqsw_starttime <pid>: kernel start time (stat field 22). It survives exec, and
|
||||
# pairing it with the pid reduces PID-reuse risk; it is not an absolute guarantee.
|
||||
sqsw_starttime() { sed 's/.*) //' "/proc/$1/stat" 2>/dev/null | cut -d' ' -f20; }
|
||||
|
||||
# sqsw_same <pid> <starttime>: the recorded process is still alive.
|
||||
sqsw_same() { [ -n "$2" ] && sqsw_alive "$1" && [ "$(sqsw_starttime "$1")" = "$2" ]; }
|
||||
|
||||
# sqsw_write_config <file>: the guest's own server. It listens only on
|
||||
# loopback, with no monitor, route or cluster listener. Its max_payload admits
|
||||
# a 1 MiB SQS message with its headers (ADR-0011:S4); the 1MB default does not.
|
||||
sqsw_write_config() {
|
||||
cat > "$1" <<CONF
|
||||
listen: "127.0.0.1:$SQSW_STORE_PORT"
|
||||
server_name: "sqs-store"
|
||||
max_payload: 2MiB
|
||||
jetstream {
|
||||
store_dir: "$SQSW_DATA/jetstream"
|
||||
sync_interval: always
|
||||
}
|
||||
CONF
|
||||
}
|
||||
|
||||
# sqsw_prepare_store: write and test the server config once boot has settled
|
||||
# the receipt key. Returns non-zero, leaving the store unstarted, otherwise.
|
||||
sqsw_prepare_store() {
|
||||
rm -f "$SQSW_RUN/store.conf"
|
||||
[ -s "$SQSW_RUN/receipt-ready" ] || { sqsw_log "store: receipt key not settled, store not prepared"; return 1; }
|
||||
umask 077
|
||||
sqsw_write_config "$SQSW_RUN/store.conf.new" || return 1
|
||||
if ! "$SQSW_NATS_BIN" -t -c "$SQSW_RUN/store.conf.new" > "$SQSW_RUN/config-test.out" 2>&1; then
|
||||
sqsw_log "store: FATAL: server config test failed: $(tr '\n' ' ' < "$SQSW_RUN/config-test.out")"
|
||||
return 1
|
||||
fi
|
||||
mv "$SQSW_RUN/store.conf.new" "$SQSW_RUN/store.conf"
|
||||
sqsw_log "store: config written for 127.0.0.1:$SQSW_STORE_PORT; server not started yet"
|
||||
}
|
||||
|
||||
# sqsw_park <why>: init respawns a step that exits, so a step that must not run
|
||||
# logs once and waits for shutdown instead of looping.
|
||||
sqsw_park() {
|
||||
sqsw_log "$1"
|
||||
exec $SQSW_PARK
|
||||
}
|
||||
|
||||
# sqsw_record <name>: record this process's identity before the stopping
|
||||
# check, so a stop that misses the record has already set stopping.
|
||||
sqsw_record() { echo "$$ $(sqsw_starttime $$)" > "$SQSW_RUN/$1.pid"; }
|
||||
|
||||
# sqsw_boot: settle the disk and receipt key once per guest boot. A transient
|
||||
# failure (exit 75) waits and exits so init runs it again; boot never remounts
|
||||
# a disk already mounted. Any other failure is final for this guest boot.
|
||||
sqsw_boot() {
|
||||
[ ! -e "$SQSW_RUN/boot-refused" ] || sqsw_park "store: boot refused, not starting"
|
||||
[ -s "$SQSW_RUN/receipt-ready" ] && return 0
|
||||
# The boot child records itself before its stopping check, as the other
|
||||
# steps do, so shutdown either sees its record or it never starts.
|
||||
SQSW_RUN="$SQSW_RUN" SQSW_BIN="$SQSW_BIN" sh -c '. "$1"; sqsw_record boot; [ ! -e "$SQSW_RUN/stopping" ] || exit 75; exec "$SQSW_BIN" boot' \
|
||||
sh "$SQSW_LIB" >> "$SQSW_CONSOLE" 2>&1 &
|
||||
child=$!
|
||||
rc=0
|
||||
wait "$child" || rc=$?
|
||||
rm -f "$SQSW_RUN/boot.pid"
|
||||
[ ! -e "$SQSW_RUN/stopping" ] || sqsw_park "store: shutdown during boot, not starting"
|
||||
case "$rc" in
|
||||
0) return 0 ;;
|
||||
75) sqsw_log "store: boot not settled, retrying in ${SQSW_BOOT_RETRY}s"
|
||||
sleep "$SQSW_BOOT_RETRY"
|
||||
exit 1 ;;
|
||||
*) : > "$SQSW_RUN/boot-refused"
|
||||
sqsw_park "store: boot refused (rc=$rc), not starting" ;;
|
||||
esac
|
||||
}
|
||||
|
||||
# sqsw_store: the respawned server body. It starts only on a prepared config.
|
||||
sqsw_store() {
|
||||
mkdir -p "$SQSW_RUN"
|
||||
sqsw_record store
|
||||
[ ! -e "$SQSW_RUN/stopping" ] || sqsw_park "store: shutdown in progress, not starting"
|
||||
sqsw_boot
|
||||
if [ ! -s "$SQSW_RUN/store.conf" ] && ! sqsw_prepare_store; then
|
||||
: > "$SQSW_RUN/boot-refused"
|
||||
sqsw_park "store: not prepared, not starting"
|
||||
fi
|
||||
sqsw_log "store: starting pid=$$"
|
||||
# ADR-0011:S11 bounds the store's Go heap; the guest must be larger.
|
||||
GOMEMLIMIT=512MiB exec "$SQSW_NATS_BIN" -c "$SQSW_RUN/store.conf"
|
||||
}
|
||||
|
||||
# sqsw_run: the respawned readiness body. A predecessor that was killed or
|
||||
# crashed could not remove its marker, so every start removes it first.
|
||||
sqsw_run() {
|
||||
mkdir -p "$SQSW_RUN"
|
||||
sqsw_record ready
|
||||
if [ -e "$SQSW_RUN/store-ready" ]; then
|
||||
rm -f "$SQSW_RUN/store-ready"
|
||||
sqsw_log "ready: removed a store-ready marker left by an earlier watcher"
|
||||
fi
|
||||
[ ! -e "$SQSW_RUN/stopping" ] || sqsw_park "ready: shutdown in progress, not starting"
|
||||
[ ! -e "$SQSW_RUN/boot-refused" ] || sqsw_park "ready: boot refused, readiness not started"
|
||||
# Boot may still be retrying; check again after a gap rather than parking.
|
||||
if [ ! -s "$SQSW_RUN/store.conf" ]; then
|
||||
sleep "$SQSW_RESPAWN_GAP"
|
||||
exit 0
|
||||
fi
|
||||
# Init respawns at once; a watcher that keeps exiting restarts at most
|
||||
# once per gap instead of in a tight loop.
|
||||
now=$(date +%s)
|
||||
last=$(cat "$SQSW_RUN/ready.last" 2>/dev/null || echo 0)
|
||||
wait=$((last + SQSW_RESPAWN_GAP - now))
|
||||
if [ "$wait" -gt 0 ] && [ "$wait" -le "$SQSW_RESPAWN_GAP" ]; then sleep "$wait"; fi
|
||||
date +%s > "$SQSW_RUN/ready.last"
|
||||
sqsw_log "ready: starting watcher pid=$$"
|
||||
exec "$SQSW_BIN" ready >> "$SQSW_CONSOLE" 2>&1
|
||||
}
|
||||
|
||||
# sqsw_end <name>: SIGTERM the recorded process, SIGKILL only after the
|
||||
# budget. Returns non-zero while it is still alive.
|
||||
sqsw_end() {
|
||||
rec=$(cat "$SQSW_RUN/$1.pid" 2>/dev/null || true)
|
||||
pid=${rec%% *} st=${rec#* }
|
||||
[ "$st" != "$rec" ] || st=""
|
||||
if [ -z "$pid" ] || ! sqsw_same "$pid" "$st"; then
|
||||
sqsw_log "stop: no running $1"
|
||||
return 0
|
||||
fi
|
||||
kill -TERM "$pid" 2>/dev/null || true
|
||||
sqsw_log "stop: SIGTERM sent to $1 pid=$pid"
|
||||
i=0
|
||||
while sqsw_same "$pid" "$st" && [ "$i" -lt "$SQSW_STOP_BUDGET_TENTHS" ]; do sleep 0.1; i=$((i + 1)); done
|
||||
if sqsw_same "$pid" "$st"; then
|
||||
sqsw_log "stop: $1 pid=$pid still running after $((SQSW_STOP_BUDGET_TENTHS / 10))s, SIGKILL"
|
||||
kill -KILL "$pid" 2>/dev/null || true
|
||||
j=0
|
||||
while sqsw_same "$pid" "$st" && [ "$j" -lt 30 ]; do sleep 0.1; j=$((j + 1)); done
|
||||
fi
|
||||
if sqsw_same "$pid" "$st"; then
|
||||
sqsw_log "stop: $1 pid=$pid STILL ALIVE"
|
||||
return 1
|
||||
fi
|
||||
sqsw_log "stop: $1 pid=$pid exited"
|
||||
rm -f "$SQSW_RUN/$1.pid"
|
||||
}
|
||||
|
||||
# sqsw_stop: block any respawn, drop the marker, stop the watcher, the server
|
||||
# and any boot in progress, and unmount the disk only once all are confirmed
|
||||
# gone. A survivor keeps the disk mounted.
|
||||
sqsw_stop() {
|
||||
mkdir -p "$SQSW_RUN"
|
||||
: > "$SQSW_RUN/stopping"
|
||||
rm -f "$SQSW_RUN/store-ready"
|
||||
if ! sqsw_end ready || ! sqsw_end store || ! sqsw_end boot; then
|
||||
rm -f "$SQSW_RUN/store-ready"
|
||||
sqsw_log "stop: leaving $SQSW_DATA mounted"
|
||||
return 1
|
||||
fi
|
||||
rm -f "$SQSW_RUN/store-ready"
|
||||
sync
|
||||
if grep -q " $SQSW_DATA " "$SQSW_MOUNTS" 2>/dev/null; then
|
||||
rc=0
|
||||
"$UMOUNT" "$SQSW_DATA" || rc=$?
|
||||
sqsw_log "stop: umount $SQSW_DATA rc=$rc"
|
||||
fi
|
||||
}
|
||||
|
||||
# sqsw_shutdown: the only shutdown action. The generic read-only remount runs
|
||||
# only after both processes stopped; a survivor keeps every mount as it is.
|
||||
sqsw_shutdown() {
|
||||
if ! sqsw_stop; then
|
||||
sync
|
||||
sqsw_log "shutdown: a guest process survived, skipping umount -a -r"
|
||||
return 1
|
||||
fi
|
||||
rc=0
|
||||
"$UMOUNT" -a -r || rc=$?
|
||||
echo "[init] shutdown: umount -a -r rc=$rc" >> "$SQSW_CONSOLE"
|
||||
}
|
||||
@@ -1,9 +0,0 @@
|
||||
#!/bin/sh
|
||||
# sysinit: load the disk drivers and clear this boot's markers. The respawned
|
||||
# store step mounts the disk, settles the receipt key and starts the store.
|
||||
. /usr/local/lib/sqs-worker/sqs-worker-lib.sh
|
||||
modprobe virtio_blk 2>/dev/null || true
|
||||
modprobe ext4 2>/dev/null || true
|
||||
mkdir -p "$SQSW_RUN"
|
||||
rm -f "$SQSW_RUN/receipt-ready" "$SQSW_RUN/store.conf" "$SQSW_RUN/boot-refused"
|
||||
exit 0
|
||||
@@ -1,5 +0,0 @@
|
||||
#!/bin/sh
|
||||
# Respawned by init: the worker, reporting ready only while the local store
|
||||
# answers. A watcher that must not run parks instead of exiting.
|
||||
. /usr/local/lib/sqs-worker/sqs-worker-lib.sh
|
||||
sqsw_run
|
||||
@@ -1,5 +0,0 @@
|
||||
#!/bin/sh
|
||||
# The only shutdown action: stop the worker, then the store, and unmount the
|
||||
# data disk, then run the generic read-only remount only if both are gone.
|
||||
. /usr/local/lib/sqs-worker/sqs-worker-lib.sh
|
||||
sqsw_shutdown
|
||||
@@ -1,5 +0,0 @@
|
||||
#!/bin/sh
|
||||
# Respawned by init: the guest's own SQS nats-server, started only on a
|
||||
# prepared config. A store that must not run parks instead of exiting.
|
||||
. /usr/local/lib/sqs-worker/sqs-worker-lib.sh
|
||||
sqsw_store
|
||||
@@ -1,4 +1,4 @@
|
||||
# spinifex-unit-version: 6
|
||||
# spinifex-unit-version: 5
|
||||
[Unit]
|
||||
Description=Spinifex Daemon (VM Orchestration)
|
||||
PartOf=spinifex.target
|
||||
@@ -15,7 +15,6 @@ Slice=spinifex-system.slice
|
||||
User=spinifex-daemon
|
||||
Group=spinifex
|
||||
ExecStartPre=/usr/local/bin/spx admin hostnet check
|
||||
ExecStartPre=-+/usr/local/bin/spx admin node-facts collect
|
||||
ExecStart=/usr/local/bin/spx service spinifex start
|
||||
Restart=on-failure
|
||||
RestartSec=5
|
||||
|
||||
@@ -1,10 +1,11 @@
|
||||
# spinifex-unit-version: 2
|
||||
# spinifex-unit-version: 1
|
||||
[Unit]
|
||||
Description=Spinifex host firewall (nftables input policy)
|
||||
Documentation=file:/etc/spinifex/firewall/spinifex.nft
|
||||
Wants=network-pre.target
|
||||
Before=network-pre.target shutdown.target
|
||||
DefaultDependencies=no
|
||||
ConditionPathExists=/etc/spinifex/firewall/peers.nft
|
||||
|
||||
[Service]
|
||||
Type=oneshot
|
||||
@@ -12,11 +13,8 @@ RemainAfterExit=yes
|
||||
StandardInput=null
|
||||
ProtectSystem=full
|
||||
ProtectHome=true
|
||||
# Unconditional: connection tracking loads on every node, so a flow that opens
|
||||
# while the policy is not armed survives the arm. The policy itself still loads
|
||||
# only where the daemon has written a peer file.
|
||||
ExecStart=/usr/local/lib/spinifex/spinifex-firewall-apply boot
|
||||
ExecReload=/usr/local/lib/spinifex/spinifex-firewall-apply boot
|
||||
ExecStart=/usr/local/lib/spinifex/spinifex-firewall-apply
|
||||
ExecReload=/usr/local/lib/spinifex/spinifex-firewall-apply
|
||||
TimeoutStopSec=10s
|
||||
|
||||
# No ExecStop: flushing on stop would take vpcd's MASQUERADE and per-EIP FORWARD
|
||||
|
||||
@@ -0,0 +1,57 @@
|
||||
# spinifex-unit-version: 1
|
||||
[Unit]
|
||||
Description=Spinifex SQS Message Store (dedicated JetStream)
|
||||
PartOf=spinifex.target
|
||||
ConditionPathExists=/etc/spinifex/sqs-store/sqs-store.conf
|
||||
|
||||
[Service]
|
||||
EnvironmentFile=-/etc/spinifex/telemetry.env
|
||||
Type=simple
|
||||
Slice=spinifex-system.slice
|
||||
User=spinifex-sqsstore
|
||||
Group=spinifex
|
||||
ExecStart=/usr/local/bin/spx service sqs-store start
|
||||
Restart=on-failure
|
||||
RestartSec=5
|
||||
# JetStream store shutdown and raft stepdown have no internal deadline.
|
||||
TimeoutStopSec=20
|
||||
OOMScoreAdjust=-500
|
||||
# Soft Go heap limit for the store; MemoryHigh/Max bound a runaway on top.
|
||||
Environment=GOMEMLIMIT=512MiB
|
||||
MemoryHigh=10%
|
||||
MemoryMax=20%
|
||||
Environment=SPINIFEX_CONFIG_PATH=/etc/spinifex/sqs-store/sqs-store.conf
|
||||
|
||||
# Filesystem access
|
||||
NoNewPrivileges=yes
|
||||
ProtectSystem=strict
|
||||
PrivateDevices=yes
|
||||
ReadOnlyPaths=/etc/spinifex/sqs-store /etc/spinifex/server.pem /etc/spinifex/server.key /etc/spinifex/ca.pem
|
||||
ReadWritePaths=/var/lib/spinifex/sqs-store /var/log/spinifex
|
||||
|
||||
# Hardening
|
||||
ProtectHome=yes
|
||||
PrivateTmp=yes
|
||||
RestrictSUIDSGID=yes
|
||||
LockPersonality=yes
|
||||
RestrictRealtime=yes
|
||||
RestrictNamespaces=yes
|
||||
ProtectClock=yes
|
||||
ProtectKernelTunables=yes
|
||||
ProtectKernelModules=yes
|
||||
ProtectKernelLogs=yes
|
||||
ProtectControlGroups=yes
|
||||
ProtectProc=invisible
|
||||
SystemCallArchitectures=native
|
||||
MemoryDenyWriteExecute=yes
|
||||
RestrictAddressFamilies=AF_INET AF_INET6 AF_UNIX AF_NETLINK
|
||||
CapabilityBoundingSet=
|
||||
ProtectHostname=yes
|
||||
KeyringMode=private
|
||||
RemoveIPC=yes
|
||||
LimitCORE=0
|
||||
LimitNOFILE=65536
|
||||
LimitNPROC=4096
|
||||
|
||||
[Install]
|
||||
WantedBy=spinifex.target
|
||||
@@ -1,9 +1,9 @@
|
||||
# spinifex-unit-version: 4
|
||||
# spinifex-unit-version: 3
|
||||
[Unit]
|
||||
Description=Spinifex Infrastructure Platform
|
||||
After=network-online.target
|
||||
Wants=network-online.target
|
||||
Wants=spinifex-nats.service spinifex-predastore.service spinifex-northstar.service spinifex-viperblock.service spinifex-daemon.service spinifex-awsgw.service spinifex-vpcd.service spinifex-ui.service spinifex-qmp-collector.service spinifex-shutdown.service spinifex-nats-watchdog.timer ovsdb-server.service ovs-vswitchd.service ovn-controller.service
|
||||
Wants=spinifex-nats.service spinifex-sqs-store.service spinifex-predastore.service spinifex-northstar.service spinifex-viperblock.service spinifex-daemon.service spinifex-awsgw.service spinifex-vpcd.service spinifex-ui.service spinifex-qmp-collector.service spinifex-shutdown.service spinifex-nats-watchdog.timer ovsdb-server.service ovs-vswitchd.service ovn-controller.service
|
||||
|
||||
[Install]
|
||||
WantedBy=multi-user.target
|
||||
|
||||
@@ -4,21 +4,12 @@ import (
|
||||
"flag"
|
||||
"fmt"
|
||||
"os"
|
||||
"slices"
|
||||
|
||||
"github.com/mulgadc/predastore/s3api"
|
||||
"github.com/mulgadc/spinifex/internal/awsmodel"
|
||||
acmawsapi "github.com/mulgadc/spinifex/spinifex/domains/acm/awsapi"
|
||||
ec2awsapi "github.com/mulgadc/spinifex/spinifex/domains/ec2/awsapi"
|
||||
awsapi "github.com/mulgadc/spinifex/spinifex/domains/ecr/awsapi"
|
||||
ecsawsapi "github.com/mulgadc/spinifex/spinifex/domains/ecs/awsapi"
|
||||
eksawsapi "github.com/mulgadc/spinifex/spinifex/domains/eks/awsapi"
|
||||
elbv2awsapi "github.com/mulgadc/spinifex/spinifex/domains/elbv2/awsapi"
|
||||
iamawsapi "github.com/mulgadc/spinifex/spinifex/domains/iam/awsapi"
|
||||
rdsawsapi "github.com/mulgadc/spinifex/spinifex/domains/rds/awsapi"
|
||||
route53awsapi "github.com/mulgadc/spinifex/spinifex/domains/route53/awsapi"
|
||||
sqsawsapi "github.com/mulgadc/spinifex/spinifex/domains/sqs/awsapi"
|
||||
stsawsapi "github.com/mulgadc/spinifex/spinifex/domains/sts/awsapi"
|
||||
"github.com/mulgadc/spinifex/spinifex/gateway"
|
||||
"github.com/mulgadc/spinifex/spinifex/ingress/aws/dispatch"
|
||||
|
||||
_ "github.com/mulgadc/bluebottle/pkg/fipsboot"
|
||||
@@ -57,26 +48,10 @@ func fail(err error) {
|
||||
func compareAll() ([]awsmodel.OperationCoverage, error) {
|
||||
// Registered services contribute their declared inventory, which needs no
|
||||
// live dependencies; nothing here serves a request.
|
||||
inventories := dispatch.AWSOperationInventory(map[string]dispatch.Inventory{
|
||||
awsapi.ServiceName: awsapi.OperationInventory(),
|
||||
acmawsapi.ServiceName: acmawsapi.OperationInventory(),
|
||||
route53awsapi.ServiceName: route53awsapi.OperationInventory(),
|
||||
elbv2awsapi.InventoryName: elbv2awsapi.OperationInventory(),
|
||||
rdsawsapi.ServiceName: rdsawsapi.OperationInventory(),
|
||||
ec2awsapi.ServiceName: ec2awsapi.OperationInventory(),
|
||||
ecsawsapi.ServiceName: ecsawsapi.OperationInventory(),
|
||||
eksawsapi.ServiceName: eksawsapi.OperationInventory(),
|
||||
iamawsapi.ServiceName: iamawsapi.OperationInventory(),
|
||||
sqsawsapi.ServiceName: sqsawsapi.OperationInventory(),
|
||||
stsawsapi.ServiceName: stsawsapi.OperationInventory(),
|
||||
inventories := gateway.AWSOperationInventory(map[string]dispatch.Inventory{
|
||||
awsapi.ServiceName: awsapi.OperationInventory(),
|
||||
acmawsapi.ServiceName: acmawsapi.OperationInventory(),
|
||||
})
|
||||
// The loop below walks the embedded models, so a served service with no
|
||||
// model would otherwise drop out of the report without a word.
|
||||
for name := range inventories {
|
||||
if !slices.Contains(awsmodel.Services(), awsmodel.Service(name)) {
|
||||
return nil, fmt.Errorf("no embedded AWS model for served service %s", name)
|
||||
}
|
||||
}
|
||||
coverages := make([]awsmodel.OperationCoverage, 0, len(awsmodel.Services()))
|
||||
for _, service := range awsmodel.Services() {
|
||||
inventory, ok := inventories[string(service)]
|
||||
|
||||
@@ -1,50 +0,0 @@
|
||||
// eks-etcd-member removes a dead peer from the embedded etcd cluster of an EKS
|
||||
// K3s control-plane VM. A member that is already gone exits 0; any other
|
||||
// failure exits non-zero.
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"flag"
|
||||
"fmt"
|
||||
"io"
|
||||
"os"
|
||||
"time"
|
||||
|
||||
_ "github.com/mulgadc/bluebottle/pkg/fipsboot"
|
||||
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/eks/etcdmember"
|
||||
)
|
||||
|
||||
func main() {
|
||||
os.Exit(run(os.Args[1:], os.Stdout, os.Stderr))
|
||||
}
|
||||
|
||||
func run(args []string, stdout, stderr io.Writer) int {
|
||||
if len(args) < 1 || args[0] != "remove" {
|
||||
fmt.Fprintln(stderr, "usage: eks-etcd-member remove --peer-ip 10.0.0.12 [--self-ip 10.0.0.10]")
|
||||
return 2
|
||||
}
|
||||
fs := flag.NewFlagSet("remove", flag.ContinueOnError)
|
||||
fs.SetOutput(stderr)
|
||||
var cfg etcdmember.Config
|
||||
endpoint := fs.String("endpoint", etcdmember.DefaultEndpoint, "etcd client endpoint")
|
||||
certDir := fs.String("cert-dir", etcdmember.DefaultCertDir, "directory holding the K3s etcd client.crt, client.key and server-ca.crt")
|
||||
timeout := fs.Duration("timeout", 30*time.Second, "overall deadline")
|
||||
fs.StringVar(&cfg.PeerIP, "peer-ip", "", "peer URL host of the member to remove")
|
||||
fs.StringVar(&cfg.SelfIP, "self-ip", "", "this host's own address; removal of it is refused")
|
||||
if err := fs.Parse(args[1:]); err != nil {
|
||||
return 2
|
||||
}
|
||||
|
||||
ctx, cancel := context.WithTimeout(context.Background(), *timeout)
|
||||
defer cancel()
|
||||
|
||||
cli, err := etcdmember.Dial(*endpoint, *certDir, *timeout)
|
||||
if err != nil {
|
||||
fmt.Fprintln(stderr, "etcd member remove failed:", err)
|
||||
return 1
|
||||
}
|
||||
defer func() { _ = cli.Close() }()
|
||||
return etcdmember.Run(ctx, cli, cfg, stdout)
|
||||
}
|
||||
@@ -1,63 +0,0 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"testing"
|
||||
|
||||
"github.com/mulgadc/spinifex/internal/testkit/fakeetcd"
|
||||
"github.com/stretchr/testify/assert"
|
||||
)
|
||||
|
||||
var fakeMembers = []fakeetcd.Member{
|
||||
{ID: 0xa1, Name: "cp-a", PeerURLs: []string{"https://10.0.0.10:2380"}},
|
||||
{ID: 0xb2, Name: "cp-b", PeerURLs: []string{"https://10.0.0.12:2380"}},
|
||||
}
|
||||
|
||||
func TestRunRequiresTheRemoveSubcommand(t *testing.T) {
|
||||
t.Parallel()
|
||||
for _, args := range [][]string{nil, {"list"}} {
|
||||
var out, errOut bytes.Buffer
|
||||
assert.Equal(t, 2, run(args, &out, &errOut))
|
||||
assert.Contains(t, errOut.String(), "usage: eks-etcd-member remove")
|
||||
}
|
||||
}
|
||||
|
||||
func TestRunRejectsUnknownFlags(t *testing.T) {
|
||||
t.Parallel()
|
||||
var out, errOut bytes.Buffer
|
||||
assert.Equal(t, 2, run([]string{"remove", "--bogus"}, &out, &errOut))
|
||||
}
|
||||
|
||||
func TestRunFailsWhenEtcdCertificatesAreMissing(t *testing.T) {
|
||||
t.Parallel()
|
||||
var out, errOut bytes.Buffer
|
||||
code := run([]string{"remove", "--peer-ip", "10.0.0.12", "--cert-dir", t.TempDir()}, &out, &errOut)
|
||||
assert.Equal(t, 1, code)
|
||||
assert.Contains(t, errOut.String(), "etcd member remove failed")
|
||||
}
|
||||
|
||||
func TestRunRemovesTheNamedPeer(t *testing.T) {
|
||||
t.Parallel()
|
||||
srv := fakeetcd.Start(t, 0xa1, fakeMembers)
|
||||
var out, errOut bytes.Buffer
|
||||
|
||||
code := run([]string{"remove", "--peer-ip", "10.0.0.12", "--self-ip", "10.0.0.10",
|
||||
"--endpoint", srv.Endpoint, "--cert-dir", srv.CertDir}, &out, &errOut)
|
||||
|
||||
assert.Zero(t, code, errOut.String()+out.String())
|
||||
assert.Contains(t, out.String(), "removed etcd member 10.0.0.12")
|
||||
assert.Equal(t, []uint64{0xb2}, srv.Removed())
|
||||
}
|
||||
|
||||
func TestRunTreatsAnAbsentPeerAsSuccess(t *testing.T) {
|
||||
t.Parallel()
|
||||
srv := fakeetcd.Start(t, 0xa1, fakeMembers)
|
||||
var out, errOut bytes.Buffer
|
||||
|
||||
code := run([]string{"remove", "--peer-ip", "10.0.0.99",
|
||||
"--endpoint", srv.Endpoint, "--cert-dir", srv.CertDir}, &out, &errOut)
|
||||
|
||||
assert.Zero(t, code)
|
||||
assert.Contains(t, out.String(), "member already absent")
|
||||
assert.Empty(t, srv.Removed())
|
||||
}
|
||||
@@ -6,7 +6,6 @@
|
||||
//
|
||||
// eks-gateway-fetch -resource addons # GET /clusters/{cluster}/internal-addons/{accountId}
|
||||
// eks-gateway-fetch -resource recovery -instance-id i-… # GET /clusters/{cluster}/internal-recovery/{accountId}/{instanceId}
|
||||
// eks-gateway-fetch -resource workers # GET /clusters/{cluster}/internal-workers/{accountId}
|
||||
//
|
||||
// Flags default to environment variables seeded by cloud-init:
|
||||
// EKS_GATEWAY_URL, EKS_GATEWAY_CA, EKS_ACCESS_KEY, EKS_SECRET_KEY, EKS_REGION,
|
||||
@@ -34,7 +33,7 @@ func main() {
|
||||
flag.StringVar(&cfg.AccountID, "account-id", os.Getenv("EKS_ACCOUNT_ID"), "Cluster account ID")
|
||||
flag.StringVar(&cfg.ClusterName, "cluster", os.Getenv("EKS_CLUSTER_NAME"), "Cluster name")
|
||||
flag.StringVar(&cfg.InstanceID, "instance-id", os.Getenv("EKS_INSTANCE_ID"), "This member's EC2 instance ID (recovery resource)")
|
||||
flag.StringVar(&cfg.Resource, "resource", "addons", "Resource to fetch: addons | recovery | workers")
|
||||
flag.StringVar(&cfg.Resource, "resource", "addons", "Resource to fetch: addons | recovery")
|
||||
flag.Parse()
|
||||
|
||||
if err := gatewayfetch.Run(context.Background(), cfg); err != nil {
|
||||
|
||||
@@ -1,32 +0,0 @@
|
||||
// Command eks-pod-identity-webhook is the IRSA mutating admission webhook run
|
||||
// on every EKS control-plane server; the implementation is in
|
||||
// agents/eks/podidentity. It listens on loopback only and is called by the
|
||||
// local kube-apiserver.
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"log/slog"
|
||||
"os"
|
||||
"os/signal"
|
||||
"syscall"
|
||||
|
||||
_ "github.com/mulgadc/bluebottle/pkg/fipsboot"
|
||||
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/eks/podidentity"
|
||||
)
|
||||
|
||||
func main() {
|
||||
log := slog.New(slog.NewTextHandler(os.Stderr, nil))
|
||||
cfg, err := podidentity.ParseServiceConfig(os.Args[1:], os.Getenv, os.Stderr)
|
||||
if err != nil {
|
||||
log.Error("invalid configuration", "error", err)
|
||||
os.Exit(2)
|
||||
}
|
||||
ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM)
|
||||
defer stop()
|
||||
if err := podidentity.Run(ctx, cfg, log); err != nil {
|
||||
log.Error("pod identity webhook stopped", "error", err)
|
||||
os.Exit(1)
|
||||
}
|
||||
}
|
||||
@@ -29,7 +29,6 @@ func loadConfig() (tokenwebhook.Config, error) {
|
||||
CertPath: envOr("EKS_WEBHOOK_CERT", "/etc/spinifex-eks/token-webhook.crt"),
|
||||
KeyPath: envOr("EKS_WEBHOOK_KEY", "/etc/spinifex-eks/token-webhook.key"),
|
||||
Kubeconfig: envOr("EKS_WEBHOOK_KUBECONFIG", "/etc/spinifex-eks/token-webhook.kubeconfig"),
|
||||
ReadyAddr: os.Getenv("EKS_WEBHOOK_READY_ADDR"),
|
||||
}
|
||||
// Static SigV4 creds are optional: when absent, eksgw.New signs with the
|
||||
// AWS SDK chain (IMDS instance-role creds), the same path the CP VM's
|
||||
|
||||
@@ -1,63 +0,0 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"crypto/tls"
|
||||
"errors"
|
||||
"fmt"
|
||||
"os"
|
||||
|
||||
lambdav1 "github.com/mulgadc/spinifex/contracts/lambda/v1"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/hostlink"
|
||||
)
|
||||
|
||||
// Exit statuses: the guest refused to start, the control connection failed,
|
||||
// or the host closed it.
|
||||
const (
|
||||
exitRefused = 2
|
||||
exitLink = 1
|
||||
exitClosed = 0
|
||||
)
|
||||
|
||||
type startup struct {
|
||||
fwCfgDir string
|
||||
fwCfgRoot string
|
||||
unload func() error
|
||||
}
|
||||
|
||||
func defaultStartup() startup {
|
||||
if _, err := os.Stat(hostlink.ConfigDiskRoot); err == nil {
|
||||
return configDiskStartup(hostlink.ConfigDiskRoot, hostlink.ConfigDiskDevice)
|
||||
}
|
||||
return startup{fwCfgDir: hostlink.FwCfgDir, fwCfgRoot: hostlink.FwCfgRoot, unload: hostlink.UnloadFwCfg}
|
||||
}
|
||||
|
||||
// configDiskStartup reads the entries init extracted from the configuration
|
||||
// disk, and seals by erasing the disk and removing the extracted tree.
|
||||
func configDiskStartup(root, device string) startup {
|
||||
return startup{
|
||||
fwCfgDir: root + "/by_name/" + lambdav1.FwCfgPrefix,
|
||||
fwCfgRoot: root,
|
||||
unload: func() error { return hostlink.SealConfigDisk(device, root) },
|
||||
}
|
||||
}
|
||||
|
||||
// connect loads the identity, seals fw_cfg whether or not the load succeeded,
|
||||
// then dials the host. Under TLS 1.3 a host refusal is seen only on first read.
|
||||
func (s startup) connect(ctx context.Context) (*tls.Conn, lambdav1.Scope, int, error) {
|
||||
id, loadErr := hostlink.LoadIdentity(s.fwCfgDir)
|
||||
if err := hostlink.SealFwCfg(s.fwCfgRoot, s.unload); err != nil {
|
||||
return nil, "", exitRefused, errors.Join(loadErr, err)
|
||||
}
|
||||
if loadErr != nil {
|
||||
return nil, "", exitRefused, loadErr
|
||||
}
|
||||
conn, err := hostlink.Dial(ctx, id)
|
||||
if err != nil {
|
||||
if ctx.Err() != nil {
|
||||
err = fmt.Errorf("%w: %w", err, context.Cause(ctx))
|
||||
}
|
||||
return nil, "", exitLink, err
|
||||
}
|
||||
return conn, id.Scope, 0, nil
|
||||
}
|
||||
@@ -1,43 +0,0 @@
|
||||
//test:in-package — package main cannot be imported by an external test package.
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
// An identity that does not load still seals the configuration disk: the
|
||||
// guest key on it is erased before the agent refuses to start.
|
||||
func TestConfigDiskStartup_RefusedLoadErasesDisk(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
root := filepath.Join(dir, "cfg")
|
||||
require.NoError(t, os.MkdirAll(filepath.Join(root, "by_name"), 0o700))
|
||||
device := filepath.Join(dir, "vdb")
|
||||
require.NoError(t, os.WriteFile(device, bytes.Repeat([]byte("guest-key"), 512), 0o600))
|
||||
|
||||
_, _, code, err := configDiskStartup(root, device).connect(t.Context())
|
||||
require.Error(t, err)
|
||||
require.Equal(t, exitRefused, code)
|
||||
got, rerr := os.ReadFile(device)
|
||||
require.NoError(t, rerr)
|
||||
require.Len(t, got, 9*512)
|
||||
require.Equal(t, make([]byte, len(got)), got, "configuration disk erased")
|
||||
}
|
||||
|
||||
// A disk that cannot be erased fails closed: the agent refuses, and the
|
||||
// extracted tree is kept so the seal check sees it.
|
||||
func TestConfigDiskStartup_UnerasableDiskRefuses(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
root := filepath.Join(dir, "cfg")
|
||||
require.NoError(t, os.MkdirAll(filepath.Join(root, "by_name"), 0o700))
|
||||
|
||||
_, _, code, err := configDiskStartup(root, filepath.Join(dir, "missing")).connect(t.Context())
|
||||
require.ErrorContains(t, err, "erase configuration disk")
|
||||
require.Equal(t, exitRefused, code)
|
||||
require.DirExists(t, root)
|
||||
}
|
||||
@@ -1,66 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"errors"
|
||||
"fmt"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
lambdav1 "github.com/mulgadc/spinifex/contracts/lambda/v1"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/hostlink"
|
||||
)
|
||||
|
||||
const entryGrant = "test-grant"
|
||||
|
||||
// testGrant is the grant the host launched this VM under.
|
||||
type testGrant struct {
|
||||
grantID, attemptID, environmentID string
|
||||
generation uint64
|
||||
}
|
||||
|
||||
// readGrant reads the test-grant fw_cfg entry: grant ID, attempt ID,
|
||||
// environment ID and generation, one per line.
|
||||
func readGrant(fwCfgDir string) (testGrant, error) {
|
||||
b, err := os.ReadFile(filepath.Join(fwCfgDir, entryGrant, "raw"))
|
||||
if err != nil {
|
||||
return testGrant{}, fmt.Errorf("read %s: %w", entryGrant, err)
|
||||
}
|
||||
lines := strings.Split(strings.TrimSuffix(string(b), "\n"), "\n")
|
||||
if len(lines) != 4 {
|
||||
return testGrant{}, errors.New(entryGrant + " entry malformed")
|
||||
}
|
||||
g := testGrant{grantID: lines[0], attemptID: lines[1], environmentID: lines[2]}
|
||||
g.generation, err = strconv.ParseUint(lines[3], 10, 64)
|
||||
if err != nil || g.generation == 0 || g.grantID == "" || g.attemptID == "" || g.environmentID == "" {
|
||||
return testGrant{}, errors.New(entryGrant + " entry malformed")
|
||||
}
|
||||
return g, nil
|
||||
}
|
||||
|
||||
// checkBinding refuses a run unless the grant names the authenticated environment
|
||||
// generation, the request carries that grant and its deadline is still ahead. The
|
||||
// clock check is a local guard only: it cannot detect skew or a superseded grant.
|
||||
func checkBinding(scope lambdav1.Scope, g testGrant, req hostlink.Request, now time.Time) error {
|
||||
parts := strings.Split(string(scope), "/")
|
||||
if len(parts) != 5 {
|
||||
return errors.New("authenticated scope malformed")
|
||||
}
|
||||
if parts[3] != g.environmentID || parts[4] != strconv.FormatUint(g.generation, 10) {
|
||||
return errors.New("grant is not for the authenticated environment generation")
|
||||
}
|
||||
if req.GrantID != g.grantID || req.AttemptID != g.attemptID {
|
||||
return errors.New("request does not carry this VM's grant")
|
||||
}
|
||||
if req.DeadlineMs <= 0 {
|
||||
return errors.New("request has no deadline")
|
||||
}
|
||||
if !now.Before(time.UnixMilli(req.DeadlineMs)) {
|
||||
return errors.New("request deadline already passed")
|
||||
}
|
||||
return nil
|
||||
}
|
||||
@@ -1,98 +0,0 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"crypto/tls"
|
||||
"net"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
lambdav1 "github.com/mulgadc/spinifex/contracts/lambda/v1"
|
||||
"github.com/mulgadc/spinifex/spinifex/runtime/compute/lambda/guestconn"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
const scope lambdav1.Scope = "ap-southeast-2/000000000001/fn-1/env-1/1"
|
||||
|
||||
// guest is a fake fw_cfg tree; unload removes it as unloading the module would.
|
||||
type guest struct {
|
||||
root string
|
||||
unloads int
|
||||
sealedAt chan struct{}
|
||||
}
|
||||
|
||||
func newGuest(t *testing.T, entries map[string][]byte) *guest {
|
||||
t.Helper()
|
||||
g := &guest{root: filepath.Join(t.TempDir(), "qemu_fw_cfg"), sealedAt: make(chan struct{}, 1)}
|
||||
for name, b := range entries {
|
||||
dir := filepath.Join(g.root, "by_name", lambdav1.FwCfgPrefix, name)
|
||||
require.NoError(t, os.MkdirAll(dir, 0o700))
|
||||
require.NoError(t, os.WriteFile(filepath.Join(dir, "raw"), b, 0o600))
|
||||
}
|
||||
return g
|
||||
}
|
||||
|
||||
func (g *guest) startup() startup {
|
||||
return startup{
|
||||
fwCfgDir: filepath.Join(g.root, "by_name", lambdav1.FwCfgPrefix),
|
||||
fwCfgRoot: g.root,
|
||||
unload: func() error {
|
||||
g.unloads++
|
||||
g.sealedAt <- struct{}{}
|
||||
return os.RemoveAll(g.root)
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
type host struct {
|
||||
ln *net.TCPListener
|
||||
gen *guestconn.Generation
|
||||
conn chan accepted
|
||||
}
|
||||
|
||||
func newHost(t *testing.T) *host {
|
||||
t.Helper()
|
||||
ln, err := net.ListenTCP("tcp", &net.TCPAddr{IP: net.IPv4(127, 0, 0, 1)})
|
||||
require.NoError(t, err)
|
||||
t.Cleanup(func() { ln.Close() })
|
||||
gen, err := guestconn.NewGeneration(scope, time.Hour)
|
||||
require.NoError(t, err)
|
||||
return &host{ln: ln, gen: gen, conn: make(chan accepted, 1)}
|
||||
}
|
||||
|
||||
func (h *host) entries() map[string][]byte { return h.gen.GuestEntries(h.ln.Addr().String()) }
|
||||
|
||||
type accepted struct {
|
||||
conn *tls.Conn
|
||||
err error
|
||||
}
|
||||
|
||||
// accept runs the host end until ctx ends and reports what it accepted.
|
||||
func (h *host) accept(ctx context.Context) {
|
||||
go func() {
|
||||
c, err := h.gen.AcceptGuest(ctx, h.ln, 2*time.Second)
|
||||
h.conn <- accepted{c, err}
|
||||
}()
|
||||
}
|
||||
|
||||
// guest waits for the host to authenticate the guest.
|
||||
func (h *host) guest(t *testing.T) *tls.Conn {
|
||||
t.Helper()
|
||||
a := wait(t, h.conn, "host to authenticate the guest")
|
||||
require.NoError(t, a.err)
|
||||
return a.conn
|
||||
}
|
||||
|
||||
// wait bounds every test wait so a stalled host or startup fails promptly.
|
||||
func wait[T any](t *testing.T, ch <-chan T, what string) T {
|
||||
t.Helper()
|
||||
select {
|
||||
case v := <-ch:
|
||||
return v
|
||||
case <-time.After(10 * time.Second):
|
||||
t.Fatalf("timed out waiting for %s", what)
|
||||
panic("unreachable")
|
||||
}
|
||||
}
|
||||
@@ -1,24 +0,0 @@
|
||||
// lambda-agent runs inside a Lambda guest.
|
||||
//
|
||||
// A normal build reads its per-generation identity from fw_cfg, unloads
|
||||
// fw_cfg so nothing started later can read the guest key, and opens the
|
||||
// authenticated control connection to the host. It accepts no invocations and
|
||||
// never starts the function's bootstrap: the message protocol and grant
|
||||
// fencing do not exist yet. The agent never listens for the host.
|
||||
//
|
||||
// A lambdatestlink build opens the same connection, then relays one invocation
|
||||
// over a test carrier for local proofs; the carrier is not the guest protocol.
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"flag"
|
||||
"os"
|
||||
)
|
||||
|
||||
func main() { os.Exit(run()) }
|
||||
|
||||
func run() int {
|
||||
flag.Parse()
|
||||
return start(context.Background())
|
||||
}
|
||||
@@ -1,286 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"archive/zip"
|
||||
"bytes"
|
||||
"crypto/sha256"
|
||||
"encoding/hex"
|
||||
"errors"
|
||||
"fmt"
|
||||
"io"
|
||||
"io/fs"
|
||||
"os"
|
||||
"os/exec"
|
||||
"path"
|
||||
"path/filepath"
|
||||
"strconv"
|
||||
"strings"
|
||||
"time"
|
||||
"unicode"
|
||||
|
||||
"golang.org/x/sys/unix"
|
||||
)
|
||||
|
||||
// Dev-only code delivery for the test path: the host attaches the ZIP as a
|
||||
// read-only disk and passes its exact size and SHA-256 through fw_cfg. This
|
||||
// is not the production delivery, and these bounds are not AWS limits.
|
||||
const (
|
||||
entryPackageSize = "test-package-size"
|
||||
entryPackageSHA256 = "test-package-sha256"
|
||||
entryHandler = "test-handler"
|
||||
|
||||
devMaxZipBytes = 16 << 20
|
||||
devMaxUnzipBytes = 64 << 20
|
||||
devMaxEntries = 1024
|
||||
packageDevice = "/dev/vda"
|
||||
taskTmpfsSizeOpts = "mode=0755,size=72m"
|
||||
)
|
||||
|
||||
type packageSpec struct {
|
||||
size int64
|
||||
sha256 [sha256.Size]byte
|
||||
handler string
|
||||
}
|
||||
|
||||
type unpackLimits struct {
|
||||
entries int
|
||||
bytes int64
|
||||
}
|
||||
|
||||
var devLimits = unpackLimits{entries: devMaxEntries, bytes: devMaxUnzipBytes}
|
||||
|
||||
// readPackageSpec reads the expected size and SHA-256 from fw_cfg; it must
|
||||
// run before fw_cfg is sealed.
|
||||
func readPackageSpec(fwCfgDir string) (packageSpec, error) {
|
||||
var spec packageSpec
|
||||
read := func(name string) (string, error) {
|
||||
b, err := os.ReadFile(filepath.Join(fwCfgDir, name, "raw"))
|
||||
if err != nil {
|
||||
return "", fmt.Errorf("read %s: %w", name, err)
|
||||
}
|
||||
return strings.TrimSpace(string(b)), nil
|
||||
}
|
||||
s, err := read(entryPackageSize)
|
||||
if err != nil {
|
||||
return spec, err
|
||||
}
|
||||
spec.size, err = strconv.ParseInt(s, 10, 64)
|
||||
if err != nil || spec.size <= 0 || spec.size > devMaxZipBytes {
|
||||
return spec, fmt.Errorf("package size %q outside 1..%d bytes", s, devMaxZipBytes)
|
||||
}
|
||||
h, err := read(entryPackageSHA256)
|
||||
if err != nil {
|
||||
return spec, err
|
||||
}
|
||||
b, err := hex.DecodeString(h)
|
||||
if err != nil || len(b) != sha256.Size {
|
||||
return spec, errors.New("package sha256 is not 64 hex digits")
|
||||
}
|
||||
copy(spec.sha256[:], b)
|
||||
if spec.handler, err = read(entryHandler); err != nil {
|
||||
return spec, err
|
||||
}
|
||||
if spec.handler == "" || len(spec.handler) > 128 || strings.IndexFunc(spec.handler, unicode.IsSpace) >= 0 ||
|
||||
strings.ContainsRune(spec.handler, 0) {
|
||||
return spec, errors.New("handler must be 1..128 characters with no whitespace")
|
||||
}
|
||||
return spec, nil
|
||||
}
|
||||
|
||||
// taskPackage puts the package under dir: a fresh nosuid,nodev tmpfs, made
|
||||
// read-only once the files are in place.
|
||||
type taskPackage struct {
|
||||
device string
|
||||
dir string
|
||||
loadDriver func() error
|
||||
mountDir func(dir string) error
|
||||
sealDir func(dir string) error
|
||||
}
|
||||
|
||||
func devTaskPackage() taskPackage {
|
||||
return taskPackage{device: packageDevice, dir: taskRoot, loadDriver: loadBlockDriver, mountDir: mountTaskDir, sealDir: sealTaskDir}
|
||||
}
|
||||
|
||||
func (p taskPackage) install(spec packageSpec) error {
|
||||
if err := p.loadDriver(); err != nil {
|
||||
return fmt.Errorf("load block driver: %w", err)
|
||||
}
|
||||
dev, err := os.Open(p.device)
|
||||
if err != nil {
|
||||
return fmt.Errorf("open package disk: %w", err)
|
||||
}
|
||||
defer dev.Close()
|
||||
if err := p.mountDir(p.dir); err != nil {
|
||||
return fmt.Errorf("mount %s: %w", p.dir, err)
|
||||
}
|
||||
if err := unpack(dev, spec, p.dir, devLimits); err != nil {
|
||||
return err
|
||||
}
|
||||
if err := p.sealDir(p.dir); err != nil {
|
||||
return fmt.Errorf("make %s read-only: %w", p.dir, err)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// unpack reads exactly spec.size bytes once, checks their SHA-256, then
|
||||
// validates every entry before writing any of them under dir.
|
||||
func unpack(r io.ReaderAt, spec packageSpec, dir string, lim unpackLimits) error {
|
||||
buf := make([]byte, spec.size)
|
||||
if _, err := io.ReadFull(io.NewSectionReader(r, 0, spec.size), buf); err != nil {
|
||||
return fmt.Errorf("read package: %w", err)
|
||||
}
|
||||
if sha256.Sum256(buf) != spec.sha256 {
|
||||
return errors.New("package sha256 mismatch")
|
||||
}
|
||||
zr, err := zip.NewReader(bytes.NewReader(buf), spec.size)
|
||||
if err != nil {
|
||||
return fmt.Errorf("open package: %w", err)
|
||||
}
|
||||
if len(zr.File) > lim.entries {
|
||||
return fmt.Errorf("package has %d entries, limit %d", len(zr.File), lim.entries)
|
||||
}
|
||||
names := make([]string, len(zr.File))
|
||||
var declared int64
|
||||
for i, f := range zr.File {
|
||||
name, err := entryName(f.Name)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
names[i] = name
|
||||
switch mode := f.Mode(); {
|
||||
case mode.IsDir():
|
||||
case mode.IsRegular():
|
||||
// Compared against what remains, so a huge size cannot wrap the sum.
|
||||
if f.UncompressedSize64 > uint64(lim.bytes-declared) {
|
||||
return fmt.Errorf("%q: unpacked size over the %d-byte limit", f.Name, lim.bytes)
|
||||
}
|
||||
declared += int64(f.UncompressedSize64)
|
||||
default:
|
||||
return fmt.Errorf("%q: %s entries are refused", f.Name, mode.Type())
|
||||
}
|
||||
}
|
||||
|
||||
root, err := os.OpenRoot(dir)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer root.Close()
|
||||
var written int64
|
||||
for i, f := range zr.File {
|
||||
n, err := extract(root, names[i], f, lim.bytes-written)
|
||||
written += n
|
||||
if err != nil {
|
||||
return fmt.Errorf("%q: %w", f.Name, err)
|
||||
}
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// extract writes one entry, refusing it if its actual bytes exceed budget
|
||||
// whatever its header claims.
|
||||
func extract(root *os.Root, name string, f *zip.File, budget int64) (int64, error) {
|
||||
if f.Mode().IsDir() {
|
||||
return 0, mkdirs(root, name)
|
||||
}
|
||||
if d := path.Dir(name); d != "." {
|
||||
if err := mkdirs(root, d); err != nil {
|
||||
return 0, err
|
||||
}
|
||||
}
|
||||
perm := os.FileMode(0o644)
|
||||
if f.Mode()&0o111 != 0 {
|
||||
perm = 0o755
|
||||
}
|
||||
in, err := f.Open()
|
||||
if err != nil {
|
||||
return 0, err
|
||||
}
|
||||
defer in.Close()
|
||||
out, err := root.OpenFile(name, os.O_CREATE|os.O_EXCL|os.O_WRONLY, perm)
|
||||
if err != nil {
|
||||
return 0, err
|
||||
}
|
||||
// The umask applies at create; fchmod does not, so the mode is exact.
|
||||
if err := out.Chmod(perm); err != nil {
|
||||
out.Close()
|
||||
return 0, err
|
||||
}
|
||||
n, err := io.Copy(out, io.LimitReader(in, budget+1))
|
||||
if cerr := out.Close(); err == nil {
|
||||
err = cerr
|
||||
}
|
||||
if err == nil && n > budget {
|
||||
err = errors.New("unpacked bytes over the limit")
|
||||
}
|
||||
return n, err
|
||||
}
|
||||
|
||||
// mkdirs creates each component of dir as mode 0755 whatever the umask. An
|
||||
// existing component must be a real directory, never a symlink.
|
||||
func mkdirs(root *os.Root, dir string) error {
|
||||
cur := ""
|
||||
for _, part := range strings.Split(dir, "/") {
|
||||
cur = path.Join(cur, part)
|
||||
if err := root.Mkdir(cur, 0o755); err != nil {
|
||||
if !errors.Is(err, fs.ErrExist) {
|
||||
return err
|
||||
}
|
||||
fi, err := root.Lstat(cur)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
if !fi.IsDir() {
|
||||
return fmt.Errorf("%q exists and is not a directory", cur)
|
||||
}
|
||||
}
|
||||
if err := root.Chmod(cur, 0o755); err != nil {
|
||||
return err
|
||||
}
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
// entryName returns a clean relative path, refusing anything that could name
|
||||
// a location outside the task root.
|
||||
func entryName(name string) (string, error) {
|
||||
if name == "" || strings.ContainsAny(name, "\\\x00") || path.IsAbs(name) {
|
||||
return "", fmt.Errorf("%q: entry name refused", name)
|
||||
}
|
||||
c := path.Clean(name)
|
||||
if c == "." || c == ".." || strings.HasPrefix(c, "../") {
|
||||
return "", fmt.Errorf("%q: entry name refused", name)
|
||||
}
|
||||
return c, nil
|
||||
}
|
||||
|
||||
// loadBlockDriver loads virtio_blk, which the image ships but init does not
|
||||
// load, then waits briefly for the package disk to appear.
|
||||
func loadBlockDriver() error {
|
||||
if err := exec.Command("/sbin/modprobe", "virtio_blk").Run(); err != nil {
|
||||
var uts unix.Utsname
|
||||
if uerr := unix.Uname(&uts); uerr != nil {
|
||||
return uerr
|
||||
}
|
||||
ko := "/lib/modules/" + unix.ByteSliceToString(uts.Release[:]) + "/kernel/drivers/block/virtio_blk.ko"
|
||||
if err := exec.Command("/sbin/insmod", ko).Run(); err != nil {
|
||||
return fmt.Errorf("modprobe and insmod virtio_blk failed: %w", err)
|
||||
}
|
||||
}
|
||||
for range 50 {
|
||||
if _, err := os.Stat(packageDevice); err == nil {
|
||||
return nil
|
||||
}
|
||||
time.Sleep(20 * time.Millisecond)
|
||||
}
|
||||
return fmt.Errorf("%s did not appear", packageDevice)
|
||||
}
|
||||
|
||||
func mountTaskDir(dir string) error {
|
||||
return unix.Mount("tmpfs", dir, "tmpfs", unix.MS_NOSUID|unix.MS_NODEV, taskTmpfsSizeOpts)
|
||||
}
|
||||
|
||||
func sealTaskDir(dir string) error {
|
||||
return unix.Mount("", dir, "", unix.MS_REMOUNT|unix.MS_RDONLY|unix.MS_NOSUID|unix.MS_NODEV, "")
|
||||
}
|
||||
@@ -1,215 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"archive/zip"
|
||||
"bytes"
|
||||
"crypto/sha256"
|
||||
"hash/crc32"
|
||||
"io/fs"
|
||||
"math"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"sort"
|
||||
"strings"
|
||||
"syscall"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
// testZip builds a ZIP; an entry named bootstrap is marked executable.
|
||||
func testZip(t *testing.T, files map[string]string) []byte {
|
||||
t.Helper()
|
||||
var buf bytes.Buffer
|
||||
zw := zip.NewWriter(&buf)
|
||||
names := make([]string, 0, len(files))
|
||||
for n := range files {
|
||||
names = append(names, n)
|
||||
}
|
||||
sort.Strings(names)
|
||||
for _, n := range names {
|
||||
h := &zip.FileHeader{Name: n, Method: zip.Deflate}
|
||||
h.SetMode(0o644)
|
||||
if n == "bootstrap" {
|
||||
h.SetMode(0o755)
|
||||
}
|
||||
w, err := zw.CreateHeader(h)
|
||||
require.NoError(t, err)
|
||||
_, err = w.Write([]byte(files[n]))
|
||||
require.NoError(t, err)
|
||||
}
|
||||
require.NoError(t, zw.Close())
|
||||
return buf.Bytes()
|
||||
}
|
||||
|
||||
// rawZip writes entries with headers set by the caller, so a test can claim
|
||||
// sizes or modes the data does not have.
|
||||
func rawZip(t *testing.T, entries ...*zip.FileHeader) []byte {
|
||||
t.Helper()
|
||||
var buf bytes.Buffer
|
||||
zw := zip.NewWriter(&buf)
|
||||
for _, h := range entries {
|
||||
w, err := zw.CreateRaw(h)
|
||||
require.NoError(t, err)
|
||||
_, err = w.Write([]byte("x"))
|
||||
require.NoError(t, err)
|
||||
}
|
||||
require.NoError(t, zw.Close())
|
||||
return buf.Bytes()
|
||||
}
|
||||
|
||||
func specFor(b []byte) packageSpec {
|
||||
return packageSpec{size: int64(len(b)), sha256: sha256.Sum256(b)}
|
||||
}
|
||||
|
||||
// Not parallel: the umask is process-wide, so the test sets a restrictive one
|
||||
// and restores it.
|
||||
func TestUnpackInstallsFilesWithFixedModes(t *testing.T) {
|
||||
old := syscall.Umask(0o077)
|
||||
t.Cleanup(func() { syscall.Umask(old) })
|
||||
b := testZip(t, map[string]string{"bootstrap": "#!/bin/sh\n", "lib/data.txt": "data"})
|
||||
dir := t.TempDir()
|
||||
require.NoError(t, unpack(bytes.NewReader(b), specFor(b), dir, devLimits))
|
||||
|
||||
pfi, err := os.Stat(filepath.Join(dir, "lib"))
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, fs.FileMode(0o755), pfi.Mode().Perm(), "implicit parent dir")
|
||||
|
||||
fi, err := os.Stat(filepath.Join(dir, "bootstrap"))
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, fs.FileMode(0o755), fi.Mode().Perm(), "executable")
|
||||
fi, err = os.Stat(filepath.Join(dir, "lib", "data.txt"))
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, fs.FileMode(0o644), fi.Mode().Perm())
|
||||
got, err := os.ReadFile(filepath.Join(dir, "lib", "data.txt"))
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, "data", string(got))
|
||||
}
|
||||
|
||||
func TestUnpackRefusesBeforeWritingAnything(t *testing.T) {
|
||||
symlink := &zip.FileHeader{Name: "link", Method: zip.Store, CRC32: crc32.ChecksumIEEE([]byte("x")), CompressedSize64: 1, UncompressedSize64: 1}
|
||||
symlink.SetMode(fs.ModeSymlink | 0o777)
|
||||
fifo := &zip.FileHeader{Name: "pipe", Method: zip.Store, CRC32: crc32.ChecksumIEEE([]byte("x")), CompressedSize64: 1, UncompressedSize64: 1}
|
||||
fifo.SetMode(fs.ModeNamedPipe | 0o644)
|
||||
// The second size would wrap a plain running sum back under the limit.
|
||||
big := func(name string, n uint64) *zip.FileHeader {
|
||||
return &zip.FileHeader{Name: name, Method: zip.Store, CRC32: crc32.ChecksumIEEE([]byte("x")), CompressedSize64: 1, UncompressedSize64: n}
|
||||
}
|
||||
cases := map[string]struct {
|
||||
zip []byte
|
||||
want string
|
||||
}{
|
||||
"parent traversal": {testZip(t, map[string]string{"a": "ok", "../escape": "x"}), "entry name refused"},
|
||||
"nested traversal": {testZip(t, map[string]string{"a": "ok", "lib/../../escape": "x"}), "entry name refused"},
|
||||
"absolute path": {testZip(t, map[string]string{"a": "ok", "/etc/escape": "x"}), "entry name refused"},
|
||||
"backslash": {testZip(t, map[string]string{"a": "ok", `..\escape`: "x"}), "entry name refused"},
|
||||
"symlink": {rawZip(t, big("a", 1), symlink), "entries are refused"},
|
||||
"named pipe": {rawZip(t, big("a", 1), fifo), "entries are refused"},
|
||||
"declared size": {rawZip(t, big("a", devMaxUnzipBytes+1)), "over the"},
|
||||
"wrapping sizes": {rawZip(t, big("a", devMaxUnzipBytes), big("b", math.MaxUint64)), "over the"},
|
||||
}
|
||||
for name, tc := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
err := unpack(bytes.NewReader(tc.zip), specFor(tc.zip), dir, devLimits)
|
||||
assert.ErrorContains(t, err, tc.want)
|
||||
left, rerr := os.ReadDir(dir)
|
||||
require.NoError(t, rerr)
|
||||
assert.Empty(t, left, "files written before the refusal")
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func TestUnpackRefusesTooManyEntries(t *testing.T) {
|
||||
b := testZip(t, map[string]string{"a": "1", "b": "2", "c": "3"})
|
||||
err := unpack(bytes.NewReader(b), specFor(b), t.TempDir(), unpackLimits{entries: 2, bytes: devMaxUnzipBytes})
|
||||
assert.ErrorContains(t, err, "limit 2")
|
||||
}
|
||||
|
||||
// The header claims a size within the limit, but the data is longer: the
|
||||
// actual bytes are refused, whatever the header says.
|
||||
func TestUnpackRefusesBytesBeyondTheLimit(t *testing.T) {
|
||||
data := bytes.Repeat([]byte("a"), 64)
|
||||
var buf bytes.Buffer
|
||||
zw := zip.NewWriter(&buf)
|
||||
h := &zip.FileHeader{Name: "a", Method: zip.Store, CRC32: crc32.ChecksumIEEE(data), CompressedSize64: uint64(len(data)), UncompressedSize64: 8}
|
||||
w, err := zw.CreateRaw(h)
|
||||
require.NoError(t, err)
|
||||
_, err = w.Write(data)
|
||||
require.NoError(t, err)
|
||||
require.NoError(t, zw.Close())
|
||||
b := buf.Bytes()
|
||||
|
||||
dir := t.TempDir()
|
||||
err = unpack(bytes.NewReader(b), specFor(b), dir, unpackLimits{entries: 4, bytes: 16})
|
||||
require.Error(t, err)
|
||||
fi, serr := os.Stat(filepath.Join(dir, "a"))
|
||||
if serr == nil {
|
||||
assert.LessOrEqual(t, fi.Size(), int64(17))
|
||||
}
|
||||
}
|
||||
|
||||
func TestUnpackChecksSizeAndHash(t *testing.T) {
|
||||
b := testZip(t, map[string]string{"bootstrap": "x"})
|
||||
wrongHash := specFor(b)
|
||||
wrongHash.sha256[0] ^= 1
|
||||
assert.ErrorContains(t, unpack(bytes.NewReader(b), wrongHash, t.TempDir(), devLimits), "sha256 mismatch")
|
||||
|
||||
longer := specFor(b)
|
||||
longer.size++
|
||||
assert.ErrorContains(t, unpack(bytes.NewReader(b), longer, t.TempDir(), devLimits), "read package")
|
||||
}
|
||||
|
||||
func TestReadPackageSpecBounds(t *testing.T) {
|
||||
sum := "e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855"
|
||||
cases := map[string]struct {
|
||||
size, hash, handler string
|
||||
want string
|
||||
}{
|
||||
"valid": {"10", sum, "bootstrap.h", ""},
|
||||
"zero": {"0", sum, "bootstrap.h", "outside"},
|
||||
"too large": {"16777217", sum, "bootstrap.h", "outside"},
|
||||
"not hex": {"10", "zz", "bootstrap.h", "64 hex digits"},
|
||||
"short": {"10", sum[:62], "bootstrap.h", "64 hex digits"},
|
||||
"empty handler": {"10", sum, "", "handler"},
|
||||
"handler space": {"10", sum, "a b", "handler"},
|
||||
"128-char handler": {"10", sum, strings.Repeat("h", 128), ""},
|
||||
"129-char handler": {"10", sum, strings.Repeat("h", 129), "handler"},
|
||||
}
|
||||
for name, tc := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
for n, v := range map[string]string{entryPackageSize: tc.size, entryPackageSHA256: tc.hash, entryHandler: tc.handler} {
|
||||
require.NoError(t, os.MkdirAll(filepath.Join(dir, n), 0o700))
|
||||
require.NoError(t, os.WriteFile(filepath.Join(dir, n, "raw"), []byte(v), 0o600))
|
||||
}
|
||||
spec, err := readPackageSpec(dir)
|
||||
if tc.want == "" {
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, int64(10), spec.size)
|
||||
assert.Equal(t, tc.handler, spec.handler)
|
||||
return
|
||||
}
|
||||
assert.ErrorContains(t, err, tc.want)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
// A pre-existing symlink where a directory belongs is refused, not followed
|
||||
// or chmodded.
|
||||
func TestMkdirsRefusesSymlinkComponent(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
outside := t.TempDir()
|
||||
require.NoError(t, os.Chmod(outside, 0o700))
|
||||
require.NoError(t, os.Symlink(outside, filepath.Join(dir, "lib")))
|
||||
root, err := os.OpenRoot(dir)
|
||||
require.NoError(t, err)
|
||||
defer root.Close()
|
||||
assert.ErrorContains(t, mkdirs(root, "lib/sub"), "not a directory")
|
||||
fi, err := os.Stat(outside)
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, fs.FileMode(0o700), fi.Mode().Perm(), "symlink target was chmodded")
|
||||
}
|
||||
@@ -1,65 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"crypto/x509"
|
||||
"errors"
|
||||
"fmt"
|
||||
"net"
|
||||
"net/url"
|
||||
"os"
|
||||
"path/filepath"
|
||||
)
|
||||
|
||||
const (
|
||||
sdkEndpointEntry = "test-sdk-endpoint"
|
||||
sdkCAEntry = "test-sdk-ca"
|
||||
sdkBundlePath = "/run/lambda-sdk/ca-bundle.pem"
|
||||
systemBundlePath = "/etc/ssl/certs/ca-certificates.crt"
|
||||
)
|
||||
|
||||
type sdkConfig struct {
|
||||
endpoint string
|
||||
ca []byte
|
||||
}
|
||||
|
||||
// readSDKConfig reads optional public settings before the boot config is sealed.
|
||||
func readSDKConfig(dir string) (*sdkConfig, error) {
|
||||
endpoint, endpointErr := os.ReadFile(filepath.Join(dir, sdkEndpointEntry, "raw"))
|
||||
ca, caErr := os.ReadFile(filepath.Join(dir, sdkCAEntry, "raw"))
|
||||
if errors.Is(endpointErr, os.ErrNotExist) && errors.Is(caErr, os.ErrNotExist) {
|
||||
return nil, nil
|
||||
}
|
||||
if endpointErr != nil || caErr != nil {
|
||||
return nil, errors.New("incomplete SDK gateway settings")
|
||||
}
|
||||
u, err := url.Parse(string(endpoint))
|
||||
if err != nil || u.Scheme != "https" || u.User != nil || u.Path != "" || u.RawQuery != "" || u.Fragment != "" ||
|
||||
net.ParseIP(u.Hostname()) == nil || u.Port() == "" {
|
||||
return nil, errors.New("invalid SDK gateway endpoint")
|
||||
}
|
||||
if len(ca) > 64<<10 || !x509.NewCertPool().AppendCertsFromPEM(ca) {
|
||||
return nil, errors.New("invalid SDK gateway CA")
|
||||
}
|
||||
return &sdkConfig{endpoint: string(endpoint), ca: ca}, nil
|
||||
}
|
||||
|
||||
// writeBundle combines the system roots with the cluster's public root.
|
||||
func (c *sdkConfig) writeBundle(path, systemPath string) error {
|
||||
system, err := os.ReadFile(systemPath)
|
||||
if err != nil && !errors.Is(err, os.ErrNotExist) {
|
||||
return fmt.Errorf("read system CA bundle: %w", err)
|
||||
}
|
||||
if err := os.MkdirAll(filepath.Dir(path), 0o755); err != nil {
|
||||
return fmt.Errorf("create SDK CA directory: %w", err)
|
||||
}
|
||||
if err := os.Chmod(filepath.Dir(path), 0o755); err != nil {
|
||||
return fmt.Errorf("make SDK CA directory readable: %w", err)
|
||||
}
|
||||
bundle := append(append(system, '\n'), c.ca...)
|
||||
if err := os.WriteFile(path, bundle, 0o644); err != nil {
|
||||
return fmt.Errorf("write SDK CA bundle: %w", err)
|
||||
}
|
||||
return os.Chmod(path, 0o644)
|
||||
}
|
||||
@@ -1,106 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
//test:in-package exercises the unexported boot configuration reader
|
||||
package main
|
||||
|
||||
import (
|
||||
"crypto/ecdsa"
|
||||
"crypto/elliptic"
|
||||
"crypto/rand"
|
||||
"crypto/tls"
|
||||
"crypto/x509"
|
||||
"crypto/x509/pkix"
|
||||
"encoding/pem"
|
||||
"math/big"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
func TestSDKConfig_OptionalAndInvalid(t *testing.T) {
|
||||
server := httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { w.WriteHeader(http.StatusOK) }))
|
||||
defer server.Close()
|
||||
ca := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: server.Certificate().Raw})
|
||||
for _, tc := range []struct {
|
||||
name string
|
||||
endpoint []byte
|
||||
ca []byte
|
||||
wantErr bool
|
||||
want bool
|
||||
}{
|
||||
{name: "absent"},
|
||||
{name: "valid", endpoint: []byte(server.URL), ca: ca, want: true},
|
||||
{name: "missing ca", endpoint: []byte(server.URL), wantErr: true},
|
||||
{name: "missing endpoint", ca: ca, wantErr: true},
|
||||
{name: "invalid endpoint", endpoint: []byte("http://127.0.0.1:1234"), ca: ca, wantErr: true},
|
||||
{name: "invalid ca", endpoint: []byte(server.URL), ca: []byte("not a certificate"), wantErr: true},
|
||||
} {
|
||||
t.Run(tc.name, func(t *testing.T) {
|
||||
dir := t.TempDir()
|
||||
for name, value := range map[string][]byte{sdkEndpointEntry: tc.endpoint, sdkCAEntry: tc.ca} {
|
||||
if value != nil {
|
||||
require.NoError(t, os.MkdirAll(filepath.Join(dir, name), 0o700))
|
||||
require.NoError(t, os.WriteFile(filepath.Join(dir, name, "raw"), value, 0o600))
|
||||
}
|
||||
}
|
||||
cfg, err := readSDKConfig(dir)
|
||||
if tc.wantErr {
|
||||
require.Error(t, err)
|
||||
require.Nil(t, cfg)
|
||||
return
|
||||
}
|
||||
require.NoError(t, err)
|
||||
if !tc.want {
|
||||
require.Nil(t, cfg)
|
||||
return
|
||||
}
|
||||
bundle := filepath.Join(t.TempDir(), "sdk", "ca.pem")
|
||||
require.NoError(t, cfg.writeBundle(bundle, filepath.Join(t.TempDir(), "absent")))
|
||||
data, err := os.ReadFile(bundle)
|
||||
require.NoError(t, err)
|
||||
roots := x509.NewCertPool()
|
||||
require.True(t, roots.AppendCertsFromPEM(data))
|
||||
client := &http.Client{Transport: &http.Transport{TLSClientConfig: &tls.Config{RootCAs: roots}}}
|
||||
resp, err := client.Get(cfg.endpoint)
|
||||
require.NoError(t, err)
|
||||
defer resp.Body.Close()
|
||||
assert.Equal(t, http.StatusOK, resp.StatusCode)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func TestSDKConfig_BundleKeepsSystemRoots(t *testing.T) {
|
||||
cluster := httptest.NewTLSServer(nil)
|
||||
defer cluster.Close()
|
||||
clusterCA := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: cluster.Certificate().Raw})
|
||||
key, err := ecdsa.GenerateKey(elliptic.P256(), rand.Reader)
|
||||
require.NoError(t, err)
|
||||
template := &x509.Certificate{SerialNumber: big.NewInt(1), Subject: pkix.Name{CommonName: "test system root"},
|
||||
NotBefore: time.Now().Add(-time.Minute), NotAfter: time.Now().Add(time.Hour),
|
||||
IsCA: true, BasicConstraintsValid: true, KeyUsage: x509.KeyUsageCertSign}
|
||||
der, err := x509.CreateCertificate(rand.Reader, template, template, &key.PublicKey, key)
|
||||
require.NoError(t, err)
|
||||
systemCA := pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: der})
|
||||
dir := t.TempDir()
|
||||
systemPath, bundlePath := filepath.Join(dir, "system.pem"), filepath.Join(dir, "sdk", "ca.pem")
|
||||
require.NoError(t, os.WriteFile(systemPath, systemCA, 0o600))
|
||||
require.NoError(t, (&sdkConfig{ca: clusterCA}).writeBundle(bundlePath, systemPath))
|
||||
bundle, err := os.ReadFile(bundlePath)
|
||||
require.NoError(t, err)
|
||||
roots := x509.NewCertPool()
|
||||
require.True(t, roots.AppendCertsFromPEM(bundle))
|
||||
client := &http.Client{Transport: &http.Transport{TLSClientConfig: &tls.Config{RootCAs: roots}}}
|
||||
resp, err := client.Get(cluster.URL)
|
||||
require.NoError(t, err)
|
||||
require.NoError(t, resp.Body.Close())
|
||||
systemCert, err := x509.ParseCertificate(der)
|
||||
require.NoError(t, err)
|
||||
_, err = systemCert.Verify(x509.VerifyOptions{Roots: roots})
|
||||
require.NoError(t, err)
|
||||
}
|
||||
@@ -1,56 +0,0 @@
|
||||
//go:build !lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"crypto/tls"
|
||||
"errors"
|
||||
"fmt"
|
||||
"io"
|
||||
"log/slog"
|
||||
)
|
||||
|
||||
// errNoProtocol is returned when the host sends anything: this build defines no
|
||||
// message protocol, so it accepts no work.
|
||||
var errNoProtocol = errors.New("host sent data, but this build accepts no invocations")
|
||||
|
||||
func start(ctx context.Context) int {
|
||||
code, err := defaultStartup().run(ctx)
|
||||
if err != nil {
|
||||
slog.Error("lambda-agent stopping", "exit", code, "err", err)
|
||||
}
|
||||
return code
|
||||
}
|
||||
|
||||
// run authenticates to the host and holds the connection until it closes.
|
||||
func (s startup) run(ctx context.Context) (int, error) {
|
||||
conn, _, code, err := s.connect(ctx)
|
||||
if err != nil {
|
||||
return code, err
|
||||
}
|
||||
defer conn.Close()
|
||||
slog.Info("control connection open; invocations are not enabled in this build", "host", conn.RemoteAddr().String())
|
||||
if err := awaitHostClose(ctx, conn); err != nil {
|
||||
return exitLink, err
|
||||
}
|
||||
return exitClosed, nil
|
||||
}
|
||||
|
||||
// awaitHostClose reads until the host closes the connection. Under TLS 1.3 a
|
||||
// host that rejects the guest certificate is only seen here, on first read.
|
||||
func awaitHostClose(ctx context.Context, conn *tls.Conn) error {
|
||||
stop := context.AfterFunc(ctx, func() { _ = conn.Close() })
|
||||
defer stop()
|
||||
n, err := conn.Read(make([]byte, 1))
|
||||
switch {
|
||||
case n > 0:
|
||||
return errNoProtocol
|
||||
case errors.Is(err, io.EOF):
|
||||
return nil
|
||||
case ctx.Err() != nil:
|
||||
return fmt.Errorf("control connection: %w", context.Cause(ctx))
|
||||
default:
|
||||
return fmt.Errorf("control connection: %w", err)
|
||||
}
|
||||
}
|
||||
@@ -1,171 +0,0 @@
|
||||
//go:build !lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"net"
|
||||
"os"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
lambdav1 "github.com/mulgadc/spinifex/contracts/lambda/v1"
|
||||
"github.com/mulgadc/spinifex/spinifex/runtime/compute/lambda/guestconn"
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
type result struct {
|
||||
code int
|
||||
err error
|
||||
}
|
||||
|
||||
func runAsync(ctx context.Context, s startup) <-chan result {
|
||||
out := make(chan result, 1)
|
||||
go func() {
|
||||
code, err := s.run(ctx)
|
||||
out <- result{code, err}
|
||||
}()
|
||||
return out
|
||||
}
|
||||
|
||||
func TestStartupAuthenticatesAfterSealingFwCfg(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
g := newGuest(t, h.entries())
|
||||
done := runAsync(ctx, g.startup())
|
||||
|
||||
conn := h.guest(t)
|
||||
select {
|
||||
case <-g.sealedAt:
|
||||
default:
|
||||
t.Fatal("guest dialled before fw_cfg was sealed")
|
||||
}
|
||||
_, err := os.Stat(g.root)
|
||||
assert.ErrorIs(t, err, os.ErrNotExist, "fw_cfg still readable while connected")
|
||||
|
||||
require.NoError(t, conn.Close())
|
||||
res := wait(t, done, "startup to return")
|
||||
assert.Equal(t, exitClosed, res.code)
|
||||
assert.NoError(t, res.err)
|
||||
assert.Equal(t, 1, g.unloads)
|
||||
}
|
||||
|
||||
func TestStartupRefusesHostData(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
done := runAsync(ctx, newGuest(t, h.entries()).startup())
|
||||
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
_, err := conn.Write([]byte{0})
|
||||
require.NoError(t, err)
|
||||
res := wait(t, done, "startup to return")
|
||||
assert.Equal(t, exitLink, res.code)
|
||||
assert.ErrorIs(t, res.err, errNoProtocol)
|
||||
}
|
||||
|
||||
// The host pins another generation's guest; the guest learns of the refusal
|
||||
// on its first read and fails with a link error, never a clean close.
|
||||
func TestStartupFailsWhenHostRefusesGuest(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
other, err := guestconn.NewGeneration(scope, time.Hour)
|
||||
require.NoError(t, err)
|
||||
entries := h.entries()
|
||||
wrong := other.GuestEntries(h.ln.Addr().String())
|
||||
entries[lambdav1.EntryGuestKey], entries[lambdav1.EntryGuestCert] = wrong[lambdav1.EntryGuestKey], wrong[lambdav1.EntryGuestCert]
|
||||
g := newGuest(t, entries)
|
||||
|
||||
res := wait(t, runAsync(ctx, g.startup()), "startup to return")
|
||||
assert.Equal(t, exitLink, res.code)
|
||||
require.Error(t, res.err)
|
||||
assert.Contains(t, res.err.Error(), "bad certificate")
|
||||
assert.Equal(t, 1, g.unloads)
|
||||
select {
|
||||
case a := <-h.conn:
|
||||
require.Error(t, a.err, "host accepted a guest it does not pin")
|
||||
default:
|
||||
}
|
||||
}
|
||||
|
||||
func TestStartupRefusesWithoutDialling(t *testing.T) {
|
||||
h := newHost(t)
|
||||
cases := map[string]struct {
|
||||
entries map[string][]byte
|
||||
unload func(g *guest) func() error
|
||||
want string
|
||||
}{
|
||||
"missing entry": {
|
||||
entries: func() map[string][]byte { e := h.entries(); delete(e, lambdav1.EntryHostCert); return e }(),
|
||||
want: "read host-cert",
|
||||
},
|
||||
"fw_cfg stays readable": {
|
||||
entries: h.entries(),
|
||||
unload: func(*guest) func() error { return func() error { return errors.New("module is built in") } },
|
||||
want: "fw_cfg still readable",
|
||||
},
|
||||
"unload claims success but tree remains": {
|
||||
entries: h.entries(),
|
||||
unload: func(*guest) func() error { return func() error { return nil } },
|
||||
want: "unload reported success",
|
||||
},
|
||||
}
|
||||
for name, tc := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
g := newGuest(t, tc.entries)
|
||||
s := g.startup()
|
||||
if tc.unload != nil {
|
||||
s.unload = tc.unload(g)
|
||||
}
|
||||
code, err := s.run(context.Background())
|
||||
assert.Equal(t, exitRefused, code)
|
||||
assert.ErrorContains(t, err, tc.want)
|
||||
})
|
||||
}
|
||||
|
||||
// Nothing dialled the host in any case above.
|
||||
require.NoError(t, h.ln.SetDeadline(time.Now().Add(50*time.Millisecond)))
|
||||
_, err := h.ln.Accept()
|
||||
var ne net.Error
|
||||
require.ErrorAs(t, err, &ne)
|
||||
assert.True(t, ne.Timeout())
|
||||
}
|
||||
|
||||
// A missing entry still seals fw_cfg: the key may be present even when
|
||||
// another entry is not.
|
||||
func TestStartupSealsEvenWhenIdentityIsIncomplete(t *testing.T) {
|
||||
h := newHost(t)
|
||||
e := h.entries()
|
||||
delete(e, lambdav1.EntryHostAddr)
|
||||
g := newGuest(t, e)
|
||||
code, err := g.startup().run(context.Background())
|
||||
assert.Equal(t, exitRefused, code)
|
||||
assert.ErrorContains(t, err, "read host-addr")
|
||||
assert.Equal(t, 1, g.unloads)
|
||||
_, statErr := os.Stat(g.root)
|
||||
assert.ErrorIs(t, statErr, os.ErrNotExist)
|
||||
}
|
||||
|
||||
func TestStartupStopsWithContext(t *testing.T) {
|
||||
ctx, cancel := context.WithCancelCause(context.Background())
|
||||
defer cancel(nil)
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
done := runAsync(ctx, newGuest(t, h.entries()).startup())
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
|
||||
reason := errors.New("guest shutting down")
|
||||
cancel(reason)
|
||||
res := wait(t, done, "startup to return")
|
||||
assert.Equal(t, exitLink, res.code)
|
||||
assert.ErrorIs(t, res.err, reason)
|
||||
}
|
||||
@@ -1,231 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"io"
|
||||
"log/slog"
|
||||
"net"
|
||||
"net/http"
|
||||
"os"
|
||||
"os/exec"
|
||||
"sort"
|
||||
"strings"
|
||||
"time"
|
||||
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/agent"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/hostlink"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/runtimeapi"
|
||||
"github.com/mulgadc/spinifex/spinifex/foundation/aws/lambdaenv"
|
||||
)
|
||||
|
||||
const (
|
||||
runtimeAddr = "127.0.0.1:9001"
|
||||
taskRoot = "/var/task"
|
||||
)
|
||||
|
||||
// testlink relays one invocation over the test carrier on the authenticated
|
||||
// connection; the carrier is not the guest protocol.
|
||||
type testlink struct {
|
||||
startup
|
||||
pkg taskPackage
|
||||
runtimeAddr string
|
||||
startBootstrap func(ctx context.Context, env []string) (*exec.Cmd, error)
|
||||
caBundlePath string
|
||||
systemCAPath string
|
||||
}
|
||||
|
||||
func start(ctx context.Context) int {
|
||||
t := testlink{
|
||||
startup: defaultStartup(),
|
||||
pkg: devTaskPackage(),
|
||||
runtimeAddr: runtimeAddr,
|
||||
caBundlePath: sdkBundlePath,
|
||||
systemCAPath: systemBundlePath,
|
||||
startBootstrap: func(ctx context.Context, env []string) (*exec.Cmd, error) {
|
||||
return agent.StartBootstrap(ctx, taskRoot+"/bootstrap", env, os.Stdout, os.Stderr)
|
||||
},
|
||||
}
|
||||
code, err := t.run(ctx)
|
||||
if err != nil {
|
||||
slog.Error("lambda-agent stopping", "exit", code, "err", err)
|
||||
}
|
||||
return code
|
||||
}
|
||||
|
||||
func (t testlink) run(parent context.Context) (int, error) {
|
||||
// The package values are fw_cfg entries, so read them before connect seals it.
|
||||
spec, specErr := readPackageSpec(t.fwCfgDir)
|
||||
grant, grantErr := readGrant(t.fwCfgDir)
|
||||
sdk, sdkErr := readSDKConfig(t.fwCfgDir)
|
||||
conn, scope, code, err := t.connect(parent)
|
||||
if err != nil {
|
||||
return code, errors.Join(err, specErr, grantErr)
|
||||
}
|
||||
defer conn.Close()
|
||||
if err := errors.Join(specErr, grantErr); err != nil {
|
||||
return exitRefused, err
|
||||
}
|
||||
link := hostlink.NewTestStream(conn)
|
||||
|
||||
// The handshake can return before the host refuses the guest, so nothing
|
||||
// tenant-facing starts until the host's first valid frame arrives.
|
||||
first, err := link.Next(parent)
|
||||
if errors.Is(err, hostlink.ErrShutdown) {
|
||||
return exitClosed, nil
|
||||
}
|
||||
if err != nil {
|
||||
return exitLink, fmt.Errorf("first frame from host: %w", err)
|
||||
}
|
||||
if parent.Err() != nil {
|
||||
return exitLink, context.Cause(parent)
|
||||
}
|
||||
// Nothing of the tenant's is installed or run for an unbound request.
|
||||
if err := checkBinding(scope, grant, first, time.Now()); err != nil {
|
||||
return exitRefused, err
|
||||
}
|
||||
// A host that sent no credential is refused before anything is installed.
|
||||
if first.Credential == nil || !first.Credential.Complete() {
|
||||
return exitRefused, errors.New("first frame carries no execution credential")
|
||||
}
|
||||
cred := *first.Credential
|
||||
first.Credential = nil
|
||||
if err := lambdaenv.Validate(cred.Environment); err != nil {
|
||||
return exitRefused, errors.New("first frame carries an invalid function environment")
|
||||
}
|
||||
if sdkErr != nil {
|
||||
slog.Warn("lambda-agent: SDK gateway settings unavailable", "err", sdkErr)
|
||||
}
|
||||
if sdk != nil {
|
||||
if err := sdk.writeBundle(t.caBundlePath, t.systemCAPath); err != nil {
|
||||
slog.Warn("lambda-agent: SDK gateway settings unavailable", "err", err)
|
||||
sdk = nil
|
||||
}
|
||||
}
|
||||
if err := t.pkg.install(spec); err != nil {
|
||||
return exitRefused, fmt.Errorf("install package: %w", err)
|
||||
}
|
||||
|
||||
srv := runtimeapi.New()
|
||||
ln, err := net.Listen("tcp", t.runtimeAddr)
|
||||
if err != nil {
|
||||
return exitLink, fmt.Errorf("listen runtime api: %w", err)
|
||||
}
|
||||
hs := &http.Server{Handler: srv.Handler(), ReadHeaderTimeout: 10 * time.Second}
|
||||
go func() { _ = hs.Serve(ln) }()
|
||||
defer hs.Close()
|
||||
|
||||
ctx, cancel := context.WithCancelCause(parent)
|
||||
defer cancel(nil)
|
||||
region, _, _ := strings.Cut(string(scope), "/")
|
||||
cmd, err := t.startBootstrap(ctx, bootstrapEnv(ln.Addr().String(), spec.handler, cred, region, sdk, t.caBundlePath))
|
||||
if err != nil {
|
||||
return exitLink, fmt.Errorf("start bootstrap: %w", err)
|
||||
}
|
||||
exited := make(chan struct{})
|
||||
go func() {
|
||||
defer close(exited)
|
||||
err := cmd.Wait()
|
||||
// Once a response is being received the result stands, so only an
|
||||
// exit before that cancels the invocation.
|
||||
if srv.BootstrapExited() {
|
||||
cancel(fmt.Errorf("bootstrap exited: %w", err))
|
||||
}
|
||||
}()
|
||||
|
||||
err = agent.Run(ctx, &oneInvocation{Link: link, first: &first}, srv)
|
||||
cancel(errors.New("agent stopping"))
|
||||
<-exited
|
||||
if err != nil {
|
||||
return exitLink, err
|
||||
}
|
||||
return exitClosed, nil
|
||||
}
|
||||
|
||||
// bootstrapEnv supplies the runtime, execution-role and SDK settings.
|
||||
func bootstrapEnv(runtimeAPI, handler string, cred hostlink.Credential, region string, sdk *sdkConfig, caPath string) []string {
|
||||
vars := map[string]string{
|
||||
"AWS_LAMBDA_RUNTIME_API": runtimeAPI,
|
||||
"_HANDLER": handler,
|
||||
"LAMBDA_TASK_ROOT": taskRoot,
|
||||
"PATH": "/bin:/sbin:/usr/bin",
|
||||
"AWS_ACCESS_KEY_ID": cred.AccessKeyID,
|
||||
"AWS_SECRET_ACCESS_KEY": cred.SecretAccessKey,
|
||||
"AWS_SESSION_TOKEN": cred.SessionToken,
|
||||
"AWS_REGION": region,
|
||||
"AWS_DEFAULT_REGION": region,
|
||||
}
|
||||
if sdk != nil {
|
||||
vars["AWS_ENDPOINT_URL"], vars["AWS_CA_BUNDLE"] = sdk.endpoint, caPath
|
||||
}
|
||||
for k, v := range cred.Environment {
|
||||
vars[k] = v
|
||||
}
|
||||
keys := make([]string, 0, len(vars))
|
||||
for k := range vars {
|
||||
keys = append(keys, k)
|
||||
}
|
||||
sort.Strings(keys)
|
||||
env := make([]string, 0, len(keys))
|
||||
for _, k := range keys {
|
||||
env = append(env, k+"="+vars[k])
|
||||
}
|
||||
return env
|
||||
}
|
||||
|
||||
// shutdownGrace is how long past the invocation deadline the agent waits for
|
||||
// the host's shutdown frame once its result is written.
|
||||
const shutdownGrace = 2 * time.Second
|
||||
|
||||
// oneInvocation serves the host's first request and then stops the agent:
|
||||
// each test VM runs a single invocation. After the result is written it waits
|
||||
// for the host's shutdown frame or EOF, so the result is not cut off by the
|
||||
// poweroff that follows the agent's exit.
|
||||
type oneInvocation struct {
|
||||
hostlink.Link
|
||||
first *hostlink.Request
|
||||
deadlineMs int64
|
||||
}
|
||||
|
||||
var _ hostlink.Link = (*oneInvocation)(nil)
|
||||
|
||||
func (o *oneInvocation) Next(ctx context.Context) (hostlink.Request, error) {
|
||||
if err := ctx.Err(); err != nil {
|
||||
return hostlink.Request{}, err
|
||||
}
|
||||
if o.first != nil {
|
||||
req := *o.first
|
||||
o.first = nil
|
||||
o.deadlineMs = req.DeadlineMs
|
||||
return req, nil
|
||||
}
|
||||
return hostlink.Request{}, o.awaitShutdown(ctx)
|
||||
}
|
||||
|
||||
// awaitShutdown reads the host's next frame and always ends the agent: with
|
||||
// ErrShutdown for a shutdown frame, EOF, a stray frame or an expired wait.
|
||||
func (o *oneInvocation) awaitShutdown(ctx context.Context) error {
|
||||
end := time.UnixMilli(o.deadlineMs)
|
||||
if now := time.Now(); end.Before(now) {
|
||||
end = now
|
||||
}
|
||||
wctx, cancel := context.WithDeadline(ctx, end.Add(shutdownGrace))
|
||||
defer cancel()
|
||||
_, err := o.Link.Next(wctx)
|
||||
switch {
|
||||
case errors.Is(err, hostlink.ErrShutdown), errors.Is(err, io.EOF):
|
||||
return hostlink.ErrShutdown
|
||||
case err == nil:
|
||||
slog.Warn("lambda-agent: frame after the invocation was not dispatched")
|
||||
return hostlink.ErrShutdown
|
||||
case ctx.Err() != nil:
|
||||
return err
|
||||
case wctx.Err() != nil:
|
||||
slog.Warn("lambda-agent: no shutdown from the host; exiting")
|
||||
return hostlink.ErrShutdown
|
||||
}
|
||||
return err
|
||||
}
|
||||
@@ -1,589 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
package main
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"context"
|
||||
"crypto/sha256"
|
||||
"encoding/hex"
|
||||
"encoding/pem"
|
||||
"io"
|
||||
"net/http"
|
||||
"net/http/httptest"
|
||||
"os"
|
||||
"os/exec"
|
||||
"path/filepath"
|
||||
"strconv"
|
||||
"strings"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
lambdav1 "github.com/mulgadc/spinifex/contracts/lambda/v1"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/lambda/hostlink"
|
||||
"github.com/mulgadc/spinifex/spinifex/runtime/compute/lambda/guestconn"
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
const helperEnv = "LAMBDA_AGENT_TEST_BOOTSTRAP"
|
||||
|
||||
// helperHandlerOut names the file the helper bootstrap writes its _HANDLER to.
|
||||
const helperHandlerOut = "LAMBDA_AGENT_TEST_HANDLER_OUT"
|
||||
|
||||
// TestHelperBootstrap is the fake function bootstrap, run as a subprocess: it
|
||||
// echoes one invocation through the Runtime API, then waits for the next.
|
||||
func TestHelperBootstrap(t *testing.T) {
|
||||
if os.Getenv(helperEnv) != "1" {
|
||||
return
|
||||
}
|
||||
if out := os.Getenv(helperHandlerOut); out != "" {
|
||||
require.NoError(t, os.WriteFile(out, []byte(os.Getenv("_HANDLER")), 0o600))
|
||||
}
|
||||
base := "http://" + os.Getenv("AWS_LAMBDA_RUNTIME_API") + "/2018-06-01/runtime/invocation/"
|
||||
resp, err := http.Get(base + "next")
|
||||
require.NoError(t, err)
|
||||
body, err := io.ReadAll(resp.Body)
|
||||
require.NoError(t, err)
|
||||
resp.Body.Close()
|
||||
id := resp.Header.Get("Lambda-Runtime-Aws-Request-Id")
|
||||
post, err := http.Post(base+id+"/response", "application/json", bytes.NewReader(body))
|
||||
require.NoError(t, err)
|
||||
post.Body.Close()
|
||||
if os.Getenv(helperExitAfterPost) == "1" {
|
||||
return
|
||||
}
|
||||
_, _ = http.Get(base + "next")
|
||||
}
|
||||
|
||||
// helperExitAfterPost makes the helper bootstrap exit once its response is posted.
|
||||
const helperExitAfterPost = "LAMBDA_AGENT_TEST_EXIT_AFTER_POST"
|
||||
|
||||
// testGrantEntry grants g-1 for attempt a-1 on the fixture scope's env-1/1.
|
||||
const testGrantEntry = "g-1\na-1\nenv-1\n1\n"
|
||||
|
||||
// testHandler is the handler the fixture's host configures.
|
||||
const testHandler = "bootstrap.fixture"
|
||||
|
||||
type bootstrapStarts struct {
|
||||
n int
|
||||
started chan struct{}
|
||||
task string
|
||||
sealed bool
|
||||
// handlerOut receives the _HANDLER the bootstrap process itself saw.
|
||||
handlerOut string
|
||||
caBundle string
|
||||
// env is the environment the agent gave the bootstrap.
|
||||
env []string
|
||||
}
|
||||
|
||||
// testlink builds a guest whose fw_cfg also carries a controlled package,
|
||||
// installed into a temp dir without mounting anything.
|
||||
func (b *bootstrapStarts) testlink(t *testing.T, entries map[string][]byte) testlink {
|
||||
t.Helper()
|
||||
zipBytes := testZip(t, map[string]string{"bootstrap": "#!/bin/sh\n", "lib/data.txt": "data"})
|
||||
disk := filepath.Join(t.TempDir(), "disk")
|
||||
require.NoError(t, os.WriteFile(disk, zipBytes, 0o600))
|
||||
sum := sha256.Sum256(zipBytes)
|
||||
entries[entryPackageSize] = []byte(strconv.Itoa(len(zipBytes)))
|
||||
entries[entryPackageSHA256] = []byte(hex.EncodeToString(sum[:]))
|
||||
if _, ok := entries[entryHandler]; !ok {
|
||||
entries[entryHandler] = []byte(testHandler)
|
||||
}
|
||||
if _, ok := entries[entryGrant]; !ok {
|
||||
entries[entryGrant] = []byte(testGrantEntry)
|
||||
}
|
||||
g := newGuest(t, entries)
|
||||
b.started = make(chan struct{}, 1)
|
||||
b.task = t.TempDir()
|
||||
b.handlerOut = filepath.Join(t.TempDir(), "handler")
|
||||
b.caBundle = filepath.Join(t.TempDir(), "sdk", "ca-bundle.pem")
|
||||
return testlink{
|
||||
startup: g.startup(),
|
||||
pkg: taskPackage{
|
||||
device: disk, dir: b.task,
|
||||
loadDriver: func() error { return nil },
|
||||
mountDir: func(string) error { return nil },
|
||||
sealDir: func(string) error { b.sealed = true; return nil },
|
||||
},
|
||||
runtimeAddr: "127.0.0.1:0",
|
||||
caBundlePath: b.caBundle,
|
||||
systemCAPath: filepath.Join(t.TempDir(), "system-ca.pem"),
|
||||
startBootstrap: func(ctx context.Context, env []string) (*exec.Cmd, error) {
|
||||
b.n++
|
||||
b.env = env
|
||||
b.started <- struct{}{}
|
||||
cmd := exec.CommandContext(ctx, os.Args[0], "-test.run=^TestHelperBootstrap$")
|
||||
cmd.Env = append(env, helperEnv+"=1", helperHandlerOut+"="+b.handlerOut)
|
||||
return cmd, cmd.Start()
|
||||
},
|
||||
}
|
||||
}
|
||||
|
||||
func runTestlinkAsync(ctx context.Context, tl testlink) <-chan result {
|
||||
out := make(chan result, 1)
|
||||
go func() {
|
||||
code, err := tl.run(ctx)
|
||||
out <- result{code, err}
|
||||
}()
|
||||
return out
|
||||
}
|
||||
|
||||
type result struct {
|
||||
code int
|
||||
err error
|
||||
}
|
||||
|
||||
func TestTestlinkServesOneInvocationThenStops(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
select {
|
||||
case <-b.started:
|
||||
t.Fatal("bootstrap started before the host sent a frame")
|
||||
case <-time.After(100 * time.Millisecond):
|
||||
}
|
||||
hs := hostlink.NewTestStream(conn)
|
||||
req := hostlink.Request{Credential: &testCredential, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: []byte(`{"a":40,"b":2}`)}
|
||||
require.NoError(t, hs.HostSend(req))
|
||||
got, err := hs.HostResult(ctx)
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, hostlink.Response{GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", Payload: req.Payload}, got)
|
||||
|
||||
// A second invocation on the same VM is never served. The agent may
|
||||
// already have closed the connection, so a failed send is a refusal too.
|
||||
second := req
|
||||
second.RequestID = "r-2"
|
||||
sendErr := hs.HostSend(second)
|
||||
res := wait(t, done, "agent to stop after one invocation")
|
||||
assert.Equal(t, exitClosed, res.code)
|
||||
assert.NoError(t, res.err)
|
||||
if sendErr == nil {
|
||||
_, err = hs.HostResult(ctx)
|
||||
assert.Error(t, err, "agent answered a second invocation")
|
||||
}
|
||||
assert.Equal(t, 1, b.n)
|
||||
seen, err := os.ReadFile(b.handlerOut)
|
||||
require.NoError(t, err, "bootstrap process did not report its handler")
|
||||
assert.Equal(t, testHandler, string(seen), "bootstrap process not given the configured handler")
|
||||
assert.True(t, b.sealed, "task dir not made read-only")
|
||||
assert.FileExists(t, filepath.Join(b.task, "lib", "data.txt"))
|
||||
assert.ElementsMatch(t, []string{
|
||||
"AWS_LAMBDA_RUNTIME_API=" + envValue(b.env, "AWS_LAMBDA_RUNTIME_API"),
|
||||
"_HANDLER=" + testHandler,
|
||||
"LAMBDA_TASK_ROOT=" + taskRoot,
|
||||
"PATH=/bin:/sbin:/usr/bin",
|
||||
"AWS_ACCESS_KEY_ID=" + testCredential.AccessKeyID,
|
||||
"AWS_SECRET_ACCESS_KEY=" + testCredential.SecretAccessKey,
|
||||
"AWS_SESSION_TOKEN=" + testCredential.SessionToken,
|
||||
"AWS_REGION=ap-southeast-2",
|
||||
"AWS_DEFAULT_REGION=ap-southeast-2",
|
||||
}, b.env, "the bootstrap gets exactly the runtime and credential variables")
|
||||
assert.NotEmpty(t, envValue(b.env, "AWS_LAMBDA_RUNTIME_API"))
|
||||
}
|
||||
|
||||
func TestTestlinkProvidesSDKSettingsToBootstrap(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
server := httptest.NewTLSServer(http.HandlerFunc(func(w http.ResponseWriter, _ *http.Request) { w.WriteHeader(http.StatusOK) }))
|
||||
defer server.Close()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
entries := h.entries()
|
||||
entries[sdkEndpointEntry] = []byte(server.URL)
|
||||
entries[sdkCAEntry] = pem.EncodeToMemory(&pem.Block{Type: "CERTIFICATE", Bytes: server.Certificate().Raw})
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, entries))
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
hs := hostlink.NewTestStream(conn)
|
||||
req := hostlink.Request{Credential: &testCredential, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: []byte(`{}`)}
|
||||
require.NoError(t, hs.HostSend(req))
|
||||
_, err := hs.HostResult(ctx)
|
||||
require.NoError(t, err)
|
||||
res := wait(t, done, "agent to stop")
|
||||
require.NoError(t, res.err)
|
||||
assert.Equal(t, server.URL, envValue(b.env, "AWS_ENDPOINT_URL"))
|
||||
assert.Equal(t, b.caBundle, envValue(b.env, "AWS_CA_BUNDLE"))
|
||||
assert.Equal(t, "ap-southeast-2", envValue(b.env, "AWS_REGION"))
|
||||
assert.Equal(t, "ap-southeast-2", envValue(b.env, "AWS_DEFAULT_REGION"))
|
||||
ca, err := os.ReadFile(b.caBundle)
|
||||
require.NoError(t, err)
|
||||
assert.Contains(t, string(ca), string(entries[sdkCAEntry]))
|
||||
info, err := os.Stat(b.caBundle)
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, os.FileMode(0o644), info.Mode().Perm())
|
||||
}
|
||||
|
||||
func TestTestlinkProvidesFunctionEnvironment(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
hs := hostlink.NewTestStream(conn)
|
||||
cred := testCredential
|
||||
cred.Environment = map[string]string{"GREETING": "hello $WORLD", "PATH": "/my/bin"}
|
||||
req := hostlink.Request{Credential: &cred, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: []byte(`{}`)}
|
||||
require.NoError(t, hs.HostSend(req))
|
||||
_, err := hs.HostResult(ctx)
|
||||
require.NoError(t, err)
|
||||
require.NoError(t, wait(t, done, "agent to stop").err)
|
||||
assert.Equal(t, "hello $WORLD", envValue(b.env, "GREETING"))
|
||||
assert.Equal(t, "/my/bin", envValue(b.env, "PATH"))
|
||||
assert.Equal(t, testCredential.SecretAccessKey, envValue(b.env, "AWS_SECRET_ACCESS_KEY"))
|
||||
}
|
||||
|
||||
func TestTestlinkRejectsReservedFunctionEnvironment(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
cred := testCredential
|
||||
cred.Environment = map[string]string{"AWS_SECRET_ACCESS_KEY": "forged"}
|
||||
req := hostlink.Request{Credential: &cred, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: []byte(`{}`)}
|
||||
require.NoError(t, hostlink.NewTestStream(conn).HostSend(req))
|
||||
result := wait(t, done, "agent to refuse the environment")
|
||||
assert.Equal(t, exitRefused, result.code)
|
||||
assert.Error(t, result.err)
|
||||
assert.NotContains(t, result.err.Error(), "forged")
|
||||
assert.Zero(t, b.n)
|
||||
}
|
||||
|
||||
var testCredential = hostlink.Credential{AccessKeyID: "ASIAGUESTKEY", SecretAccessKey: "guest-secret", SessionToken: "guest-token"}
|
||||
|
||||
func envValue(env []string, name string) string {
|
||||
for _, kv := range env {
|
||||
if v, ok := strings.CutPrefix(kv, name+"="); ok {
|
||||
return v
|
||||
}
|
||||
}
|
||||
return ""
|
||||
}
|
||||
|
||||
// A first frame without a complete credential installs and runs nothing.
|
||||
func TestTestlinkFrameWithoutCredentialNeverStartsBootstrap(t *testing.T) {
|
||||
for name, cred := range map[string]*hostlink.Credential{
|
||||
"none": nil,
|
||||
"incomplete": {AccessKeyID: "ASIAGUESTKEY"},
|
||||
} {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
require.NoError(t, hostlink.NewTestStream(conn).HostSend(hostlink.Request{Credential: cred, GrantID: "g-1", AttemptID: "a-1",
|
||||
RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: []byte(`{}`)}))
|
||||
|
||||
res := wait(t, done, "agent to stop")
|
||||
assert.Equal(t, exitRefused, res.code)
|
||||
assert.ErrorContains(t, res.err, "no execution credential")
|
||||
assert.Zero(t, b.n)
|
||||
assert.False(t, b.sealed)
|
||||
assert.NoFileExists(t, filepath.Join(b.task, "bootstrap"), "package installed")
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
// A bootstrap that posts its response and exits at once does not cost the
|
||||
// host the result; the agent then waits for the host's shutdown.
|
||||
func TestTestlinkBootstrapExitingAfterResponseStillDelivers(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
tl := b.testlink(t, h.entries())
|
||||
start := tl.startBootstrap
|
||||
tl.startBootstrap = func(ctx context.Context, env []string) (*exec.Cmd, error) {
|
||||
return start(ctx, append(env, helperExitAfterPost+"=1"))
|
||||
}
|
||||
done := runTestlinkAsync(ctx, tl)
|
||||
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
hs := hostlink.NewTestStream(conn)
|
||||
payload := bytes.Repeat([]byte("0123456789"), 300_000)
|
||||
req := hostlink.Request{Credential: &testCredential, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli(), Payload: payload}
|
||||
require.NoError(t, hs.HostSend(req))
|
||||
got, err := hs.HostResult(ctx)
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, hostlink.Response{GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1", Payload: payload}, got)
|
||||
|
||||
require.NoError(t, hs.HostShutdown())
|
||||
res := wait(t, done, "agent to stop after the host's shutdown")
|
||||
assert.Equal(t, exitClosed, res.code)
|
||||
assert.NoError(t, res.err)
|
||||
}
|
||||
|
||||
// The agent stays up for the host's shutdown rather than exiting as soon as
|
||||
// its result is written, and never dispatches a stray frame.
|
||||
func TestOneInvocationAfterResult(t *testing.T) {
|
||||
first := hostlink.Request{RequestID: "r-1", DeadlineMs: time.Now().Add(time.Minute).UnixMilli()}
|
||||
cases := map[string]struct {
|
||||
next func(context.Context) (hostlink.Request, error)
|
||||
want error
|
||||
}{
|
||||
"shutdown frame": {func(context.Context) (hostlink.Request, error) { return hostlink.Request{}, hostlink.ErrShutdown }, hostlink.ErrShutdown},
|
||||
"eof": {func(context.Context) (hostlink.Request, error) { return hostlink.Request{}, io.EOF }, hostlink.ErrShutdown},
|
||||
"a second request is not dispatched": {func(context.Context) (hostlink.Request, error) {
|
||||
return hostlink.Request{RequestID: "r-2"}, nil
|
||||
}, hostlink.ErrShutdown},
|
||||
}
|
||||
for name, tc := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
o := &oneInvocation{Link: stubLink{tc.next}, first: &first}
|
||||
_, err := o.Next(context.Background())
|
||||
require.NoError(t, err)
|
||||
req, err := o.Next(context.Background())
|
||||
require.ErrorIs(t, err, tc.want)
|
||||
assert.Empty(t, req.RequestID)
|
||||
})
|
||||
}
|
||||
t.Run("host never sends shutdown", func(t *testing.T) {
|
||||
past := first
|
||||
past.DeadlineMs = time.Now().Add(-time.Second).UnixMilli()
|
||||
o := &oneInvocation{Link: stubLink{func(ctx context.Context) (hostlink.Request, error) {
|
||||
<-ctx.Done()
|
||||
return hostlink.Request{}, ctx.Err()
|
||||
}}, first: &past}
|
||||
_, err := o.Next(context.Background())
|
||||
require.NoError(t, err)
|
||||
start := time.Now()
|
||||
_, err = o.Next(context.Background())
|
||||
require.ErrorIs(t, err, hostlink.ErrShutdown)
|
||||
assert.GreaterOrEqual(t, time.Since(start), shutdownGrace-100*time.Millisecond, "waited the grace")
|
||||
})
|
||||
t.Run("parent cancel is not a shutdown", func(t *testing.T) {
|
||||
o := &oneInvocation{Link: stubLink{func(ctx context.Context) (hostlink.Request, error) {
|
||||
<-ctx.Done()
|
||||
return hostlink.Request{}, ctx.Err()
|
||||
}}, first: &first}
|
||||
_, _ = o.Next(context.Background())
|
||||
ctx, cancel := context.WithCancel(context.Background())
|
||||
cancel()
|
||||
_, err := o.Next(ctx)
|
||||
require.ErrorIs(t, err, context.Canceled)
|
||||
})
|
||||
}
|
||||
|
||||
type stubLink struct {
|
||||
next func(context.Context) (hostlink.Request, error)
|
||||
}
|
||||
|
||||
var _ hostlink.Link = stubLink{}
|
||||
|
||||
func (l stubLink) Next(ctx context.Context) (hostlink.Request, error) { return l.next(ctx) }
|
||||
|
||||
func (stubLink) Complete(context.Context, hostlink.Response) error { return nil }
|
||||
|
||||
func TestTestlinkShutdownBeforeInvocationNeverStartsBootstrap(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
require.NoError(t, hostlink.NewTestStream(conn).HostShutdown())
|
||||
res := wait(t, done, "agent to stop")
|
||||
assert.Equal(t, exitClosed, res.code)
|
||||
assert.NoError(t, res.err)
|
||||
assert.Zero(t, b.n)
|
||||
}
|
||||
|
||||
// The guest's handshake completes, but the host pins another generation's
|
||||
// guest; the refusal arrives on first read and the bootstrap never starts.
|
||||
func TestTestlinkHostRefusalNeverStartsBootstrap(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
other, err := guestconn.NewGeneration(scope, time.Hour)
|
||||
require.NoError(t, err)
|
||||
entries := h.entries()
|
||||
wrong := other.GuestEntries(h.ln.Addr().String())
|
||||
entries[lambdav1.EntryGuestKey], entries[lambdav1.EntryGuestCert] = wrong[lambdav1.EntryGuestKey], wrong[lambdav1.EntryGuestCert]
|
||||
var b bootstrapStarts
|
||||
|
||||
res := wait(t, runTestlinkAsync(ctx, b.testlink(t, entries)), "agent to stop")
|
||||
assert.Equal(t, exitLink, res.code)
|
||||
require.Error(t, res.err)
|
||||
assert.Contains(t, res.err.Error(), "bad certificate")
|
||||
assert.Zero(t, b.n)
|
||||
}
|
||||
|
||||
func TestTestlinkHostCloseBeforeFrameNeverStartsBootstrap(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, h.entries()))
|
||||
|
||||
require.NoError(t, h.guest(t).Close())
|
||||
res := wait(t, done, "agent to stop")
|
||||
assert.Equal(t, exitLink, res.code)
|
||||
assert.ErrorContains(t, res.err, "first frame from host")
|
||||
assert.Zero(t, b.n)
|
||||
}
|
||||
|
||||
// A cancelled context is reported before the buffered request is delivered,
|
||||
// and the request stays undelivered.
|
||||
func TestOneInvocationHonoursCancelledContext(t *testing.T) {
|
||||
ctx, cancel := context.WithCancel(context.Background())
|
||||
cancel()
|
||||
o := &oneInvocation{first: &hostlink.Request{RequestID: "r-1"}}
|
||||
_, err := o.Next(ctx)
|
||||
require.ErrorIs(t, err, context.Canceled)
|
||||
|
||||
req, err := o.Next(context.Background())
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, "r-1", req.RequestID)
|
||||
}
|
||||
|
||||
// A package that fails its check is refused after authentication, and the
|
||||
// bootstrap never starts.
|
||||
func TestTestlinkBadPackageNeverStartsBootstrap(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
tl := b.testlink(t, h.entries())
|
||||
tl.pkg.device = filepath.Join(t.TempDir(), "other")
|
||||
require.NoError(t, os.WriteFile(tl.pkg.device, testZip(t, map[string]string{"bootstrap": "x"}), 0o600))
|
||||
done := runTestlinkAsync(ctx, tl)
|
||||
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
require.NoError(t, hostlink.NewTestStream(conn).HostSend(hostlink.Request{Credential: &testCredential, GrantID: "g-1", AttemptID: "a-1", RequestID: "r-1",
|
||||
DeadlineMs: time.Now().Add(5 * time.Second).UnixMilli()}))
|
||||
res := wait(t, done, "agent to stop")
|
||||
assert.Equal(t, exitRefused, res.code)
|
||||
assert.ErrorContains(t, res.err, "install package")
|
||||
assert.Zero(t, b.n)
|
||||
assert.False(t, b.sealed)
|
||||
}
|
||||
|
||||
// Missing package values still seal fw_cfg, and the agent refuses before
|
||||
// waiting for any frame.
|
||||
func TestTestlinkMissingPackageSpecRefuses(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
tl := b.testlink(t, h.entries())
|
||||
require.NoError(t, os.RemoveAll(filepath.Join(tl.fwCfgDir, entryPackageSHA256)))
|
||||
|
||||
res := wait(t, runTestlinkAsync(ctx, tl), "agent to stop")
|
||||
assert.Equal(t, exitRefused, res.code)
|
||||
assert.ErrorContains(t, res.err, "read "+entryPackageSHA256)
|
||||
assert.Zero(t, b.n)
|
||||
_, err := os.Stat(tl.fwCfgRoot)
|
||||
assert.ErrorIs(t, err, os.ErrNotExist, "fw_cfg not sealed")
|
||||
}
|
||||
|
||||
// A request that does not carry this VM's grant, or a grant for another
|
||||
// environment generation than the one authenticated, installs and runs
|
||||
// nothing of the tenant's.
|
||||
func TestTestlinkUnboundRequestNeverStartsBootstrap(t *testing.T) {
|
||||
cases := map[string]struct {
|
||||
grant string
|
||||
grantID, attempt string
|
||||
deadlineMs int64
|
||||
want string
|
||||
}{
|
||||
"frame with no deadline": {testGrantEntry, "g-1", "a-1", -1, "no deadline"},
|
||||
"frame with expired deadline": {testGrantEntry, "g-1", "a-1", time.Now().Add(-time.Second).UnixMilli(), "deadline already passed"},
|
||||
"frame for another grant": {testGrantEntry, "g-2", "a-1", 0, "does not carry this VM's grant"},
|
||||
"frame for another attempt": {testGrantEntry, "g-1", "a-2", 0, "does not carry this VM's grant"},
|
||||
"frame with no grant": {testGrantEntry, "", "", 0, "does not carry this VM's grant"},
|
||||
"grant for another env": {"g-1\na-1\nenv-2\n1\n", "g-1", "a-1", 0, "not for the authenticated environment"},
|
||||
"grant for another gen": {"g-1\na-1\nenv-1\n2\n", "g-1", "a-1", 0, "not for the authenticated environment"},
|
||||
}
|
||||
for name, tc := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
entries := h.entries()
|
||||
entries[entryGrant] = []byte(tc.grant)
|
||||
done := runTestlinkAsync(ctx, b.testlink(t, entries))
|
||||
conn := h.guest(t)
|
||||
defer conn.Close()
|
||||
deadline := time.Now().Add(5 * time.Second).UnixMilli()
|
||||
switch {
|
||||
case tc.deadlineMs < 0:
|
||||
deadline = 0
|
||||
case tc.deadlineMs > 0:
|
||||
deadline = tc.deadlineMs
|
||||
}
|
||||
req := hostlink.Request{Credential: &testCredential, GrantID: tc.grantID, AttemptID: tc.attempt, RequestID: "r-1",
|
||||
DeadlineMs: deadline, Payload: []byte(`{}`)}
|
||||
require.NoError(t, hostlink.NewTestStream(conn).HostSend(req))
|
||||
|
||||
res := wait(t, done, "agent to stop")
|
||||
assert.Equal(t, exitRefused, res.code)
|
||||
assert.ErrorContains(t, res.err, tc.want)
|
||||
assert.Zero(t, b.n)
|
||||
assert.False(t, b.sealed)
|
||||
assert.NoFileExists(t, filepath.Join(b.task, "bootstrap"), "package installed")
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
// A missing or malformed grant entry refuses before any frame is awaited.
|
||||
func TestTestlinkBadGrantEntryRefuses(t *testing.T) {
|
||||
cases := map[string]*string{
|
||||
"missing": nil,
|
||||
"too few lines": ptr("g-1\na-1\nenv-1\n"),
|
||||
"generation 0": ptr("g-1\na-1\nenv-1\n0\n"),
|
||||
"empty grant": ptr("\na-1\nenv-1\n1\n"),
|
||||
}
|
||||
for name, grant := range cases {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Second)
|
||||
defer cancel()
|
||||
h := newHost(t)
|
||||
h.accept(ctx)
|
||||
var b bootstrapStarts
|
||||
entries := h.entries()
|
||||
if grant != nil {
|
||||
entries[entryGrant] = []byte(*grant)
|
||||
}
|
||||
tl := b.testlink(t, entries)
|
||||
if grant == nil {
|
||||
require.NoError(t, os.RemoveAll(filepath.Join(tl.fwCfgDir, entryGrant)))
|
||||
}
|
||||
res := wait(t, runTestlinkAsync(ctx, tl), "agent to stop")
|
||||
assert.Equal(t, exitRefused, res.code)
|
||||
assert.ErrorContains(t, res.err, entryGrant)
|
||||
assert.Zero(t, b.n)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func ptr(s string) *string { return &s }
|
||||
@@ -1,224 +0,0 @@
|
||||
// Command lambda-image-agent-tag checks whether the lambda-agent archived in a
|
||||
// built Lambda guest initramfs carries the lambdatestlink build tag. Build
|
||||
// information is recorded by the build; it is not proof of how the agent behaves.
|
||||
package main
|
||||
|
||||
import (
|
||||
"bufio"
|
||||
"bytes"
|
||||
"compress/gzip"
|
||||
"debug/buildinfo"
|
||||
"errors"
|
||||
"fmt"
|
||||
"io"
|
||||
"os"
|
||||
"slices"
|
||||
"strconv"
|
||||
"strings"
|
||||
)
|
||||
|
||||
const (
|
||||
agentPath = "usr/local/bin/lambda-agent"
|
||||
maxArchive = 1 << 30
|
||||
maxAgent = 256 << 20
|
||||
maxName = 4096
|
||||
newcHeader = 110
|
||||
trailerName = "TRAILER!!!"
|
||||
modeTypeMask = 0o170000
|
||||
modeDir = 0o040000
|
||||
modeRegular = 0o100000
|
||||
)
|
||||
|
||||
var agentParents = []string{"usr", "usr/local", "usr/local/bin"}
|
||||
|
||||
func main() {
|
||||
if len(os.Args) != 3 || (os.Args[2] != "present" && os.Args[2] != "absent") {
|
||||
fmt.Fprintln(os.Stderr, "usage: lambda-image-agent-tag <initramfs.cpio.gz> present|absent")
|
||||
os.Exit(2)
|
||||
}
|
||||
if err := run(os.Args[1], os.Args[2] == "present"); err != nil {
|
||||
fmt.Fprintf(os.Stderr, "%s: %v\n", os.Args[1], err)
|
||||
os.Exit(1)
|
||||
}
|
||||
}
|
||||
|
||||
func run(path string, want bool) error {
|
||||
fi, err := os.Lstat(path)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
if !fi.Mode().IsRegular() {
|
||||
return errors.New("not a regular file")
|
||||
}
|
||||
f, err := os.Open(path)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
defer f.Close()
|
||||
agent, err := archivedAgent(f)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
return checkTag(agent, want)
|
||||
}
|
||||
|
||||
// archivedAgent returns the single agent entry after checking listed names,
|
||||
// entry types, the agent's link count and zero-only padding after the trailer.
|
||||
// It is not a full kernel-equivalent parser; 070702 CRCs, for one, are unchecked.
|
||||
func archivedAgent(r io.Reader) ([]byte, error) {
|
||||
zr, err := gzip.NewReader(bufio.NewReader(r))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("not a readable gzip archive: %w", err)
|
||||
}
|
||||
raw, err := io.ReadAll(io.LimitReader(zr, maxArchive+1))
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("not a readable gzip archive: %w", err)
|
||||
}
|
||||
if len(raw) > maxArchive {
|
||||
return nil, fmt.Errorf("archive larger than %d bytes", maxArchive)
|
||||
}
|
||||
|
||||
seen := map[string]bool{}
|
||||
var agent []byte
|
||||
off := 0
|
||||
for {
|
||||
e, next, err := parseEntry(raw, off)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("entry at offset %d: %w", off, err)
|
||||
}
|
||||
off = next
|
||||
if e.name == trailerName {
|
||||
break
|
||||
}
|
||||
if !canonical(e.name) {
|
||||
return nil, fmt.Errorf("entry name %q is not canonical", e.name)
|
||||
}
|
||||
if seen[e.name] {
|
||||
return nil, fmt.Errorf("entry %q appears more than once", e.name)
|
||||
}
|
||||
seen[e.name] = true
|
||||
for _, p := range agentParents {
|
||||
if e.name == p && e.mode&modeTypeMask != modeDir {
|
||||
return nil, fmt.Errorf("%s is not a directory", p)
|
||||
}
|
||||
}
|
||||
if e.name == agentPath {
|
||||
for _, p := range agentParents {
|
||||
if !seen[p] {
|
||||
return nil, fmt.Errorf("%s has no %s directory entry before it", agentPath, p)
|
||||
}
|
||||
}
|
||||
if e.mode&modeTypeMask != modeRegular {
|
||||
return nil, fmt.Errorf("%s is not a regular file", agentPath)
|
||||
}
|
||||
if e.nlink != 1 {
|
||||
return nil, fmt.Errorf("%s has %d links, want 1", agentPath, e.nlink)
|
||||
}
|
||||
if len(e.data) > maxAgent {
|
||||
return nil, fmt.Errorf("%s is larger than %d bytes", agentPath, maxAgent)
|
||||
}
|
||||
agent = e.data
|
||||
}
|
||||
}
|
||||
if i := bytes.IndexFunc(raw[off:], func(r rune) bool { return r != 0 }); i >= 0 {
|
||||
return nil, fmt.Errorf("data after the archive trailer at offset %d", off+i)
|
||||
}
|
||||
if agent == nil {
|
||||
return nil, fmt.Errorf("no %s entry", agentPath)
|
||||
}
|
||||
return agent, nil
|
||||
}
|
||||
|
||||
type entry struct {
|
||||
name string
|
||||
mode uint64
|
||||
nlink uint64
|
||||
data []byte
|
||||
}
|
||||
|
||||
func parseEntry(raw []byte, off int) (entry, int, error) {
|
||||
if len(raw)-off < newcHeader {
|
||||
return entry{}, 0, errors.New("truncated header")
|
||||
}
|
||||
h := raw[off : off+newcHeader]
|
||||
if m := string(h[:6]); m != "070701" && m != "070702" {
|
||||
return entry{}, 0, fmt.Errorf("magic %q is not newc", m)
|
||||
}
|
||||
field := func(i int) (uint64, error) {
|
||||
return strconv.ParseUint(string(h[6+8*i:14+8*i]), 16, 32)
|
||||
}
|
||||
mode, err1 := field(1)
|
||||
nlink, err2 := field(4)
|
||||
size, err3 := field(6)
|
||||
nameSize, err4 := field(11)
|
||||
if err := errors.Join(err1, err2, err3, err4); err != nil {
|
||||
return entry{}, 0, fmt.Errorf("bad header field: %w", err)
|
||||
}
|
||||
if nameSize < 2 || nameSize > maxName {
|
||||
return entry{}, 0, fmt.Errorf("name size %d out of range", nameSize)
|
||||
}
|
||||
nameEnd := off + newcHeader + int(nameSize)
|
||||
dataOff := align4(nameEnd)
|
||||
if dataOff > len(raw) || size > uint64(len(raw)-dataOff) {
|
||||
return entry{}, 0, errors.New("truncated entry")
|
||||
}
|
||||
dataEnd := dataOff + int(size)
|
||||
if dataEnd > len(raw) || align4(dataEnd) > len(raw) {
|
||||
return entry{}, 0, errors.New("truncated entry")
|
||||
}
|
||||
name := raw[off+newcHeader : nameEnd]
|
||||
if name[len(name)-1] != 0 || bytes.IndexByte(name[:len(name)-1], 0) >= 0 {
|
||||
return entry{}, 0, errors.New("name is not one NUL-terminated string")
|
||||
}
|
||||
return entry{name: string(name[:len(name)-1]), mode: mode, nlink: nlink, data: raw[dataOff:dataEnd]}, align4(dataEnd), nil
|
||||
}
|
||||
|
||||
func align4(n int) int { return (n + 3) &^ 3 }
|
||||
|
||||
// canonical accepts "." and relative names with no empty, "." or ".." component.
|
||||
func canonical(name string) bool {
|
||||
if name == "." {
|
||||
return true
|
||||
}
|
||||
if name == "" || strings.HasPrefix(name, "/") {
|
||||
return false
|
||||
}
|
||||
for c := range strings.SplitSeq(name, "/") {
|
||||
if c == "" || c == "." || c == ".." {
|
||||
return false
|
||||
}
|
||||
}
|
||||
return true
|
||||
}
|
||||
|
||||
func checkTag(agent []byte, want bool) error {
|
||||
info, err := buildinfo.Read(bytes.NewReader(agent))
|
||||
if err != nil {
|
||||
return fmt.Errorf("no Go build information in %s: %w", agentPath, err)
|
||||
}
|
||||
return decide(info, want)
|
||||
}
|
||||
|
||||
// decide refuses build information with no recorded settings, since their
|
||||
// absence does not show the tag was absent.
|
||||
func decide(info *buildinfo.BuildInfo, want bool) error {
|
||||
if len(info.Settings) == 0 {
|
||||
return fmt.Errorf("%s records no build settings", agentPath)
|
||||
}
|
||||
if hasTag(info, "lambdatestlink") != want {
|
||||
return fmt.Errorf("%s: lambdatestlink build tag present=%v, want %v", agentPath, !want, want)
|
||||
}
|
||||
return nil
|
||||
}
|
||||
|
||||
func hasTag(info *buildinfo.BuildInfo, tag string) bool {
|
||||
for _, s := range info.Settings {
|
||||
if s.Key != "-tags" {
|
||||
continue
|
||||
}
|
||||
if slices.Contains(strings.Split(s.Value, ","), tag) {
|
||||
return true
|
||||
}
|
||||
}
|
||||
return false
|
||||
}
|
||||
@@ -1,154 +0,0 @@
|
||||
package main
|
||||
|
||||
import (
|
||||
"bytes"
|
||||
"compress/gzip"
|
||||
"debug/buildinfo"
|
||||
"fmt"
|
||||
"os"
|
||||
"path/filepath"
|
||||
"runtime/debug"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
type fileEntry struct {
|
||||
name string
|
||||
mode uint32
|
||||
nlink uint32
|
||||
data string
|
||||
}
|
||||
|
||||
func dir(name string) fileEntry { return fileEntry{name: name, mode: modeDir | 0o755, nlink: 2} }
|
||||
func file(name, data string) fileEntry {
|
||||
return fileEntry{name: name, mode: modeRegular | 0o755, nlink: 1, data: data}
|
||||
}
|
||||
func symlink(name, target string) fileEntry {
|
||||
return fileEntry{name: name, mode: 0o120777, nlink: 1, data: target}
|
||||
}
|
||||
|
||||
// newc writes entries and a trailer in the format the kernel unpacks.
|
||||
func newc(entries ...fileEntry) []byte {
|
||||
var b bytes.Buffer
|
||||
for i, e := range append(entries, fileEntry{name: trailerName, nlink: 1}) {
|
||||
fmt.Fprintf(&b, "070701%08X%08X%08X%08X%08X%08X%08X%08X%08X%08X%08X%08X%08X",
|
||||
i+1, e.mode, 0, 0, e.nlink, 0, len(e.data), 0, 0, 0, 0, len(e.name)+1, 0)
|
||||
b.WriteString(e.name)
|
||||
b.WriteByte(0)
|
||||
pad(&b)
|
||||
b.WriteString(e.data)
|
||||
pad(&b)
|
||||
}
|
||||
return b.Bytes()
|
||||
}
|
||||
|
||||
func pad(b *bytes.Buffer) {
|
||||
for b.Len()%4 != 0 {
|
||||
b.WriteByte(0)
|
||||
}
|
||||
}
|
||||
|
||||
func gz(parts ...[]byte) []byte {
|
||||
var b bytes.Buffer
|
||||
for _, p := range parts {
|
||||
w := gzip.NewWriter(&b)
|
||||
_, _ = w.Write(p)
|
||||
_ = w.Close()
|
||||
}
|
||||
return b.Bytes()
|
||||
}
|
||||
|
||||
func withAgent(agent fileEntry) []byte {
|
||||
return newc(dir("."), dir("usr"), dir("usr/local"), dir("usr/local/bin"), agent, file("init", "#!/bin/sh"))
|
||||
}
|
||||
|
||||
func TestArchivedAgent_ReturnsTheOneArchivedAgent(t *testing.T) {
|
||||
got, err := archivedAgent(bytes.NewReader(gz(withAgent(file(agentPath, "agent-bytes")))))
|
||||
require.NoError(t, err)
|
||||
require.Equal(t, "agent-bytes", string(got))
|
||||
|
||||
padded := append(withAgent(file(agentPath, "agent-bytes")), make([]byte, 1024)...)
|
||||
got, err = archivedAgent(bytes.NewReader(gz(padded)))
|
||||
require.NoError(t, err, "zero padding after the trailer")
|
||||
require.Equal(t, "agent-bytes", string(got))
|
||||
}
|
||||
|
||||
func TestArchivedAgent_Refuses(t *testing.T) {
|
||||
parents := []fileEntry{dir("."), dir("usr"), dir("usr/local"), dir("usr/local/bin")}
|
||||
agent := file(agentPath, "untagged")
|
||||
small := newc(append(parents, agent)...)
|
||||
second := newc(append(parents[1:], file(agentPath, "tagged"))...)
|
||||
// The second archive starts inside what would be the first one's final block.
|
||||
sameBlock := append(append([]byte{}, small...), second...)
|
||||
for len(sameBlock)%512 != 0 {
|
||||
sameBlock = append(sameBlock, 0)
|
||||
}
|
||||
|
||||
for name, archive := range map[string][]byte{
|
||||
"aliased duplicate": gz(newc(append(parents, agent, file("././"+agentPath, "tagged"))...)),
|
||||
"leading slash alias": gz(newc(append(parents, agent, file("/"+agentPath, "tagged"))...)),
|
||||
"dot-dot component": gz(newc(append(parents, agent, file("usr/local/../local/bin/lambda-agent", "x"))...)),
|
||||
"empty component": gz(newc(append(parents, agent, file("usr//local/bin/lambda-agent", "x"))...)),
|
||||
"exact duplicate": gz(newc(append(parents, agent, file(agentPath, "tagged"))...)),
|
||||
"duplicate directory": gz(newc(dir("usr"), dir("usr"), dir("usr/local"), dir("usr/local/bin"), agent)),
|
||||
"missing directory entry": gz(newc(dir("usr"), dir("usr/local/bin"), agent)),
|
||||
"directory after the agent": gz(newc(dir("usr"), dir("usr/local"), agent, dir("usr/local/bin"))),
|
||||
"parent is a symlink": gz(newc(dir("usr"), dir("usr/local"), symlink("usr/local/bin", "elsewhere"), agent)),
|
||||
"agent is a symlink": gz(newc(append(parents, symlink(agentPath, "/bin/sh"))...)),
|
||||
"hard-linked agent": gz(newc(append(parents, fileEntry{name: agentPath, mode: modeRegular | 0o755, nlink: 2, data: "x"})...)),
|
||||
"no agent": gz(newc(parents...)),
|
||||
"second archive appended": gz(small, second),
|
||||
"second archive in same block": gz(sameBlock),
|
||||
"non-zero byte after trailer": gz(append(append([]byte{}, small...), 0, 0, 1)),
|
||||
"truncated archive": gz(small[:len(small)-20]),
|
||||
"not newc": gz([]byte("not cpio at all, but long enough to fill one newc header: " + string(make([]byte, 120)))),
|
||||
"not gzip": small,
|
||||
} {
|
||||
t.Run(name, func(t *testing.T) {
|
||||
_, err := archivedAgent(bytes.NewReader(archive))
|
||||
require.Error(t, err)
|
||||
})
|
||||
}
|
||||
}
|
||||
|
||||
func TestRun_RefusesASymlinkedImagePath(t *testing.T) {
|
||||
d := t.TempDir()
|
||||
img := filepath.Join(d, "initramfs.cpio.gz")
|
||||
require.NoError(t, os.WriteFile(img, gz(withAgent(file(agentPath, "x"))), 0o644))
|
||||
require.NoError(t, os.Symlink(img, filepath.Join(d, "link")))
|
||||
require.ErrorContains(t, run(filepath.Join(d, "link"), false), "not a regular file")
|
||||
}
|
||||
|
||||
func TestHasTag_MatchesWholeTagsOnly(t *testing.T) {
|
||||
info := func(tags string) *buildinfo.BuildInfo {
|
||||
return &buildinfo.BuildInfo{Settings: []debug.BuildSetting{{Key: "-buildmode", Value: "exe"}, {Key: "-tags", Value: tags}}}
|
||||
}
|
||||
require.True(t, hasTag(info("lambdatestlink,fips140v1.0"), "lambdatestlink"))
|
||||
require.False(t, hasTag(info("fips140v1.0"), "lambdatestlink"))
|
||||
require.False(t, hasTag(info("lambdatestlinkx,xlambdatestlink"), "lambdatestlink"))
|
||||
require.False(t, hasTag(&buildinfo.BuildInfo{}, "lambdatestlink"))
|
||||
}
|
||||
|
||||
// The test binary carries real build information, so checkTag reads it and
|
||||
// refuses the opposite answer; bytes with no build information are refused.
|
||||
func TestCheckTag_ReadsRealBuildInformation(t *testing.T) {
|
||||
exe, err := os.Executable()
|
||||
require.NoError(t, err)
|
||||
raw, err := os.ReadFile(exe)
|
||||
require.NoError(t, err)
|
||||
info, err := buildinfo.Read(bytes.NewReader(raw))
|
||||
require.NoError(t, err)
|
||||
tagged := hasTag(info, "lambdatestlink")
|
||||
require.NoError(t, checkTag(raw, tagged))
|
||||
require.Error(t, checkTag(raw, !tagged))
|
||||
require.Error(t, checkTag([]byte("not a Go binary"), false))
|
||||
}
|
||||
|
||||
func TestDecide_RefusesMissingBuildSettings(t *testing.T) {
|
||||
require.ErrorContains(t, decide(&buildinfo.BuildInfo{}, false), "no build settings")
|
||||
require.ErrorContains(t, decide(&buildinfo.BuildInfo{}, true), "no build settings")
|
||||
untagged := &buildinfo.BuildInfo{Settings: []debug.BuildSetting{{Key: "-buildmode", Value: "exe"}}}
|
||||
require.NoError(t, decide(untagged, false))
|
||||
require.Error(t, decide(untagged, true))
|
||||
}
|
||||
@@ -1,26 +0,0 @@
|
||||
//go:build lambdatestlink
|
||||
|
||||
// Command lambda-image-digest prints the test-link runtime image digest of one
|
||||
// directory, selecting the boot files the same way the daemon does. The digest
|
||||
// identifies the bytes; it does not authenticate where they came from.
|
||||
package main
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"os"
|
||||
|
||||
"github.com/mulgadc/spinifex/spinifex/runtime/compute/lambda/testlink"
|
||||
)
|
||||
|
||||
func main() {
|
||||
if len(os.Args) != 2 {
|
||||
fmt.Fprintln(os.Stderr, "usage: lambda-image-digest <image-dir>")
|
||||
os.Exit(2)
|
||||
}
|
||||
digest, err := testlink.DirDigest(os.Args[1])
|
||||
if err != nil {
|
||||
fmt.Fprintln(os.Stderr, err)
|
||||
os.Exit(1)
|
||||
}
|
||||
fmt.Println(digest)
|
||||
}
|
||||
@@ -1,140 +0,0 @@
|
||||
// sqs-worker runs inside the SQS microVM, next to its own SQS NATS server.
|
||||
//
|
||||
// Usage:
|
||||
//
|
||||
// sqs-worker boot
|
||||
// sqs-worker ready
|
||||
//
|
||||
// boot selects and mounts the guest's data disk named by fw_cfg, asks the
|
||||
// gateway for its receipt key decision with its instance-role credentials,
|
||||
// then loads the recorded key, or creates one on a new volume and records its
|
||||
// fingerprint, before the guest's SQS store first starts on that disk. ready
|
||||
// then keeps checking the guest's own loopback JetStream, keeping a readiness
|
||||
// marker only while it answers, and serves SQS owner calls against it, pulling
|
||||
// them from the gateway with the same credentials.
|
||||
//
|
||||
// Serving is refused unless the gateway reports a single-node host cluster.
|
||||
// A disk whose receipt key is missing or does not match the recorded one
|
||||
// refuses to serve; recovering it needs a procedure not yet designed, and
|
||||
// nothing deletes or replaces the volume.
|
||||
package main
|
||||
|
||||
import (
|
||||
"context"
|
||||
"errors"
|
||||
"fmt"
|
||||
"os"
|
||||
"os/signal"
|
||||
"syscall"
|
||||
"time"
|
||||
|
||||
"github.com/nats-io/nats.go/jetstream"
|
||||
|
||||
"github.com/mulgadc/spinifex/internal/gwsign"
|
||||
"github.com/mulgadc/spinifex/internal/sqsworkergw"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/sqs/workerboot"
|
||||
"github.com/mulgadc/spinifex/spinifex/agents/sqs/workerkey"
|
||||
"github.com/mulgadc/spinifex/spinifex/domains/sqs/workerserve"
|
||||
"github.com/mulgadc/spinifex/spinifex/foundation/state/clustersize"
|
||||
)
|
||||
|
||||
// The cluster CA the daemon supplies; the only root the client trusts.
|
||||
const caPath = "/sys/firmware/qemu_fw_cfg/by_name/opt/spinifex/ca-cert/raw"
|
||||
|
||||
// newClaimer signs with the IMDS instance-role credentials only; there is
|
||||
// no static-key fallback.
|
||||
func newClaimer() workerboot.NewClaimer {
|
||||
return func(region, gatewayURL string) (workerkey.Claimer, error) {
|
||||
c, err := newClient(region, gatewayURL)
|
||||
if err != nil {
|
||||
return nil, err
|
||||
}
|
||||
return workerboot.GatewayClaimer{Client: c}, nil
|
||||
}
|
||||
}
|
||||
|
||||
func newClient(region, gatewayURL string) (*sqsworkergw.Client, error) {
|
||||
ca, err := os.ReadFile(caPath)
|
||||
if err != nil {
|
||||
return nil, fmt.Errorf("read cluster CA: %w", err)
|
||||
}
|
||||
return sqsworkergw.New(gatewayURL, region, ca, gwsign.NewInstanceRole())
|
||||
}
|
||||
|
||||
func main() {
|
||||
if len(os.Args) != 2 || (os.Args[1] != "boot" && os.Args[1] != "ready") {
|
||||
fmt.Fprintln(os.Stderr, "usage: sqs-worker boot|ready")
|
||||
os.Exit(2)
|
||||
}
|
||||
logf := func(msg string) { fmt.Fprintf(os.Stderr, "[sqs-worker] %s\n", msg) }
|
||||
run := boot
|
||||
if os.Args[1] == "ready" {
|
||||
run = ready
|
||||
}
|
||||
if err := run(logf); err != nil {
|
||||
if os.Args[1] == "boot" && workerboot.Transient(err) {
|
||||
logf("boot not settled, will retry: " + err.Error())
|
||||
os.Exit(exitTransient)
|
||||
}
|
||||
logf("FATAL: " + err.Error())
|
||||
os.Exit(1)
|
||||
}
|
||||
}
|
||||
|
||||
// exitTransient tells the guest's supervisor that boot may be run again.
|
||||
const exitTransient = 75
|
||||
|
||||
func config(logf func(string)) workerboot.Config {
|
||||
return workerboot.Config{
|
||||
FwCfgDir: "/sys/firmware/qemu_fw_cfg/by_name/opt/spinifex/sqs-worker",
|
||||
SysBlock: "/sys/block",
|
||||
DevDir: "/dev",
|
||||
MountInfo: "/proc/self/mountinfo",
|
||||
DataDir: "/data",
|
||||
RunDir: "/run/sqs-worker",
|
||||
Disk: workerboot.SystemDisk{E2fsck: "/sbin/e2fsck"},
|
||||
IsBlockDevice: workerboot.IsBlockDevice,
|
||||
NewClaimer: newClaimer(),
|
||||
Attempts: 60,
|
||||
Wait: 5 * time.Second,
|
||||
Log: logf,
|
||||
}
|
||||
}
|
||||
|
||||
func boot(logf func(string)) error {
|
||||
ctx, cancel := context.WithTimeout(context.Background(), 10*time.Minute)
|
||||
defer cancel()
|
||||
_, err := workerboot.Boot(ctx, config(logf))
|
||||
return err
|
||||
}
|
||||
|
||||
// ready runs until init stops it; the marker is removed on the way out.
|
||||
func ready(logf func(string)) error {
|
||||
ctx, cancel := signal.NotifyContext(context.Background(), syscall.SIGTERM, syscall.SIGINT)
|
||||
defer cancel()
|
||||
// The store is one guest, so its streams and buckets are R1.
|
||||
if err := clustersize.Declare(1); err != nil {
|
||||
return err
|
||||
}
|
||||
err := workerboot.Watch(ctx, workerboot.ReadyConfig{
|
||||
Config: config(logf),
|
||||
Checker: workerboot.NATSChecker{},
|
||||
CallTimeout: 30 * time.Second,
|
||||
Refresh: 30 * time.Second,
|
||||
Refused: 5 * time.Minute,
|
||||
Serve: &workerboot.ServeConfig{
|
||||
Connector: workerboot.NATSConnector{},
|
||||
Run: func(ctx context.Context, region, gatewayURL string, js jetstream.JetStream, receiptKey []byte) error {
|
||||
c, err := newClient(region, gatewayURL)
|
||||
if err != nil {
|
||||
return err
|
||||
}
|
||||
return workerserve.Run(ctx, c, js, receiptKey, logf)
|
||||
},
|
||||
},
|
||||
})
|
||||
if errors.Is(err, context.Canceled) {
|
||||
return nil
|
||||
}
|
||||
return err
|
||||
}
|
||||
@@ -10,7 +10,7 @@ quota state.
|
||||
a request/reply route on which a gateway, in the `awsgw-quota-seal` queue
|
||||
group, irreversibly seals one account's quota override record.
|
||||
`domains/admission/quota` owns it: `awsgw` serves it.
|
||||
`operator/accountlifecycle` is the consumer: it calls this subject to fence every
|
||||
`accountteardown` is the consumer: it calls this subject to fence every
|
||||
override writer at the quota owner before the account record is removed.
|
||||
|
||||
The request names the account in its body (`account_id`) and may carry an
|
||||
@@ -31,20 +31,10 @@ Any current gateway may answer; the leader-accepted revision is the evidence
|
||||
whichever one does, and the override bucket is opened at seal time even on a
|
||||
gateway whose quotas were disabled at startup.
|
||||
|
||||
## Seal-quota-override capability
|
||||
|
||||
`AccountQuotaOverrideSealCapabilitySubject(node)` (`admission.v1.AccountQuotaOverrideSealCapability.<node>`) is a per-node, read-only request/reply route on which `operator/accountlifecycle` asks one configured gateway whether it supports the quota override seal, immediately before every seal attempt.
|
||||
The quota owner (`domains/admission/quota`) owns it: `awsgw` serves it with no queue group, so a request addressed to one node reaches only subscribers configured as that node, and two gateways sharing a node ID both answer.
|
||||
The consumer's quota seal reaper probes every configured node that runs `awsgw` and sends no seal request unless every one answers compatible.
|
||||
|
||||
`AccountQuotaOverrideSealCapabilityVersion` is a claim about the whole binary's writers, not about the responder: version 1 means every quota override writer in the binary is revision-guarded and refuses a sealed record, and the binary serves the seal subject.
|
||||
A gateway binary without this responder has no subscriber, so the probe for that node reads as incomplete (`no-reply`), the same as a stopped or partitioned node.
|
||||
The reply proves nothing beyond the moment it was sent: a gateway downgraded or replaced after a successful probe is outside the guarantee.
|
||||
|
||||
### Compatibility boundary
|
||||
|
||||
The subject name carries the contract major; `account_quota_seal_test.go` and `account_quota_seal_capability_test.go`
|
||||
preserve the version-1 JSON shapes and the subject builder.
|
||||
The subject name carries the contract major; `account_quota_seal_test.go`
|
||||
preserves the version-1 JSON shape.
|
||||
Producer and consumer are in the coordinated Spinifex release, so the window
|
||||
is N and N-1.
|
||||
A gateway that predates the subject has no subscription, so the request gets
|
||||
|
||||
@@ -1,39 +0,0 @@
|
||||
package admissionv1
|
||||
|
||||
import "strings"
|
||||
|
||||
// AccountQuotaOverrideSealCapabilityVersion is the version this binary reports.
|
||||
// Version 1: every quota override writer in the binary is revision-guarded and
|
||||
// refuses a sealed record, and the seal subject is served.
|
||||
const AccountQuotaOverrideSealCapabilityVersion = 1
|
||||
|
||||
// AccountQuotaOverrideSealCapabilitySubject is one gateway's read-only
|
||||
// seal-capability route, addressed per node so a probe reaches exactly the
|
||||
// node it asks about.
|
||||
func AccountQuotaOverrideSealCapabilitySubject(node string) string {
|
||||
return "admission.v1.AccountQuotaOverrideSealCapability." + node
|
||||
}
|
||||
|
||||
// ValidSealCapabilityNode reports whether node is exactly one NATS subject
|
||||
// token. Responder and probe both check it before building a subject, so
|
||||
// neither can address a wildcard or another node's route.
|
||||
func ValidSealCapabilityNode(node string) bool {
|
||||
return node != "" && !strings.ContainsAny(node, ".*>\t\n\r ")
|
||||
}
|
||||
|
||||
// AccountQuotaOverrideSealCapabilityRequest asks node whether it supports the
|
||||
// quota override seal. CorrelationRef is an opaque caller reference echoed for
|
||||
// logs; it grants nothing and is never a fence or authority to write.
|
||||
type AccountQuotaOverrideSealCapabilityRequest struct {
|
||||
Node string `json:"node"`
|
||||
CorrelationRef string `json:"correlation_ref,omitempty"`
|
||||
}
|
||||
|
||||
// AccountQuotaOverrideSealCapabilityReply reports the responding node's
|
||||
// capability version. It carries no account data and proves nothing beyond
|
||||
// the moment it was sent.
|
||||
type AccountQuotaOverrideSealCapabilityReply struct {
|
||||
Node string `json:"node"`
|
||||
CapabilityVersion int `json:"capability_version"`
|
||||
CorrelationRef string `json:"correlation_ref,omitempty"`
|
||||
}
|
||||
@@ -1,43 +0,0 @@
|
||||
package admissionv1
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
func TestAccountQuotaOverrideSealCapabilitySubject(t *testing.T) {
|
||||
require.Equal(t, "admission.v1.AccountQuotaOverrideSealCapability.node-1", AccountQuotaOverrideSealCapabilitySubject("node-1"))
|
||||
require.Equal(t, "admission.v1.AccountQuotaOverrideSealCapability.node-2", AccountQuotaOverrideSealCapabilitySubject("node-2"))
|
||||
}
|
||||
|
||||
func TestAccountQuotaOverrideSealCapabilityRequestJSONContract(t *testing.T) {
|
||||
data, err := json.Marshal(AccountQuotaOverrideSealCapabilityRequest{Node: "node-1", CorrelationRef: "req-1"})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"node":"node-1","correlation_ref":"req-1"}`, string(data))
|
||||
|
||||
data, err = json.Marshal(AccountQuotaOverrideSealCapabilityRequest{Node: "node-1"})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"node":"node-1"}`, string(data))
|
||||
}
|
||||
|
||||
func TestAccountQuotaOverrideSealCapabilityReplyJSONContract(t *testing.T) {
|
||||
reply := AccountQuotaOverrideSealCapabilityReply{Node: "node-1", CapabilityVersion: 1, CorrelationRef: "req-1"}
|
||||
data, err := json.Marshal(reply)
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"node":"node-1","capability_version":1,"correlation_ref":"req-1"}`, string(data))
|
||||
|
||||
data, err = json.Marshal(AccountQuotaOverrideSealCapabilityReply{Node: "node-1", CapabilityVersion: 1})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"node":"node-1","capability_version":1}`, string(data))
|
||||
}
|
||||
|
||||
func TestValidSealCapabilityNode(t *testing.T) {
|
||||
for _, node := range []string{"node-1", "node_2", "n3"} {
|
||||
require.True(t, ValidSealCapabilityNode(node), node)
|
||||
}
|
||||
for _, node := range []string{"", "a.b", "*", ">", "node-*", "node >", "node\t1", "node\n1"} {
|
||||
require.False(t, ValidSealCapabilityNode(node), "%q", node)
|
||||
}
|
||||
}
|
||||
@@ -1,255 +0,0 @@
|
||||
package clusterv1
|
||||
|
||||
import "regexp"
|
||||
|
||||
const (
|
||||
// NodeFactsSubject fans out a request for each daemon's hardware and
|
||||
// cabling description. Its reply payload is NodeFacts.
|
||||
NodeFactsSubject = "spinifex.node.facts"
|
||||
|
||||
nodeNetTestPrefix = "spinifex.node.nettest."
|
||||
)
|
||||
|
||||
var nodeIDPattern = regexp.MustCompile(`^[A-Za-z0-9_-]{1,64}$`)
|
||||
|
||||
// ValidNodeID reports whether a node name is safe to place in a subject: no
|
||||
// NATS token separator or wildcard can get through.
|
||||
func ValidNodeID(node string) bool {
|
||||
return nodeIDPattern.MatchString(node)
|
||||
}
|
||||
|
||||
// NodeFactsOfSubject asks one node alone for its NodeFacts.
|
||||
func NodeFactsOfSubject(node string) string {
|
||||
return NodeFactsSubject + "." + node
|
||||
}
|
||||
|
||||
// NetTestServeSubject asks one node to start a one-shot throughput server.
|
||||
func NetTestServeSubject(node string) string {
|
||||
return nodeNetTestPrefix + node + ".serve"
|
||||
}
|
||||
|
||||
// NetTestRunSubject asks one node to run the client side of a throughput test.
|
||||
func NetTestRunSubject(node string) string {
|
||||
return nodeNetTestPrefix + node + ".run"
|
||||
}
|
||||
|
||||
// Planes a node's NICs can carry, named as the installer names them.
|
||||
const (
|
||||
PlaneWAN = "wan"
|
||||
PlaneLAN = "lan"
|
||||
PlaneVPC = "vpc"
|
||||
)
|
||||
|
||||
// NetTestPort is the TCP port reserved for node-to-node throughput tests.
|
||||
const NetTestPort = 5001
|
||||
|
||||
// NodeFacts describes one node's hardware, OS and cabling. Fields a host
|
||||
// cannot report are left empty, and Sources says why.
|
||||
type NodeFacts struct {
|
||||
Node string `json:"node"`
|
||||
CollectedAt string `json:"collected_at,omitempty"`
|
||||
FormFactor string `json:"form_factor"`
|
||||
Virt string `json:"virt,omitempty"`
|
||||
System SystemFacts `json:"system"`
|
||||
OS OSFacts `json:"os"`
|
||||
CPU CPUFacts `json:"cpu"`
|
||||
Memory MemoryFacts `json:"memory"`
|
||||
NICs []NICFacts `json:"nics"`
|
||||
Planes []PlaneFacts `json:"planes"`
|
||||
Disks []DiskFacts `json:"disks,omitempty"`
|
||||
Filesystems []FilesystemFacts `json:"filesystems,omitempty"`
|
||||
Sources map[string]string `json:"sources"`
|
||||
}
|
||||
|
||||
// SystemFacts is the machine's identity from DMI.
|
||||
type SystemFacts struct {
|
||||
Vendor string `json:"vendor,omitempty"`
|
||||
Product string `json:"product,omitempty"`
|
||||
Family string `json:"family,omitempty"`
|
||||
Version string `json:"version,omitempty"`
|
||||
Serial string `json:"serial,omitempty"`
|
||||
UUID string `json:"uuid,omitempty"`
|
||||
ChassisType string `json:"chassis_type,omitempty"`
|
||||
ChassisAsset string `json:"chassis_asset,omitempty"`
|
||||
BoardVendor string `json:"board_vendor,omitempty"`
|
||||
BoardProduct string `json:"board_product,omitempty"`
|
||||
BoardSerial string `json:"board_serial,omitempty"`
|
||||
BIOSVendor string `json:"bios_vendor,omitempty"`
|
||||
BIOSVersion string `json:"bios_version,omitempty"`
|
||||
BIOSDate string `json:"bios_date,omitempty"`
|
||||
DeviceModel string `json:"device_model,omitempty"`
|
||||
}
|
||||
|
||||
// OSFacts names the running OS and Spinifex build.
|
||||
type OSFacts struct {
|
||||
Name string `json:"name,omitempty"`
|
||||
Kernel string `json:"kernel,omitempty"`
|
||||
Arch string `json:"arch,omitempty"`
|
||||
Version string `json:"spx_version,omitempty"`
|
||||
Commit string `json:"spx_commit,omitempty"`
|
||||
UptimeSeconds int64 `json:"uptime_seconds,omitempty"`
|
||||
Hostname string `json:"hostname,omitempty"`
|
||||
}
|
||||
|
||||
// CPUFacts summarises the sockets. Threads and Cores are totals. Features are
|
||||
// the flags worth naming; Flags is every flag the CPU reports.
|
||||
type CPUFacts struct {
|
||||
Model string `json:"model,omitempty"`
|
||||
Vendor string `json:"vendor,omitempty"`
|
||||
Sockets int `json:"sockets"`
|
||||
CoresPerSocket int `json:"cores_per_socket,omitempty"`
|
||||
Cores int `json:"cores"`
|
||||
Threads int `json:"threads"`
|
||||
MaxMHz int `json:"max_mhz,omitempty"`
|
||||
CurrentMHz int `json:"current_mhz,omitempty"`
|
||||
Virtualization string `json:"virtualization,omitempty"`
|
||||
Features []string `json:"features,omitempty"`
|
||||
Flags []string `json:"flags,omitempty"`
|
||||
}
|
||||
|
||||
// MemoryFacts groups identical DIMMs so a 32-slot board reads as one line.
|
||||
type MemoryFacts struct {
|
||||
TotalBytes uint64 `json:"total_bytes"`
|
||||
SlotsUsed int `json:"slots_used"`
|
||||
SlotsTotal int `json:"slots_total"`
|
||||
MaxCapacity string `json:"max_capacity,omitempty"`
|
||||
ECC string `json:"ecc,omitempty"`
|
||||
Groups []MemoryGroup `json:"groups"`
|
||||
OSTotalBytes uint64 `json:"os_total_bytes,omitempty"`
|
||||
OSAvailBytes uint64 `json:"os_available_bytes,omitempty"`
|
||||
HugePageBytes uint64 `json:"huge_page_bytes,omitempty"`
|
||||
}
|
||||
|
||||
// MemoryGroup is a set of DIMMs alike in every field shown. ConfiguredMTs
|
||||
// below SpeedMTs means the board runs them slower than rated.
|
||||
type MemoryGroup struct {
|
||||
Count int `json:"count"`
|
||||
SizeBytes uint64 `json:"size_bytes"`
|
||||
Type string `json:"type,omitempty"`
|
||||
TypeDetail string `json:"type_detail,omitempty"`
|
||||
FormFactor string `json:"form_factor,omitempty"`
|
||||
SpeedMTs int `json:"speed_mts,omitempty"`
|
||||
ConfiguredMTs int `json:"configured_mts,omitempty"`
|
||||
Manufacturer string `json:"manufacturer,omitempty"`
|
||||
PartNumber string `json:"part_number,omitempty"`
|
||||
Rank string `json:"rank,omitempty"`
|
||||
Serials []string `json:"serials,omitempty"`
|
||||
}
|
||||
|
||||
// DiskFacts is one physical drive as sysfs reports it. Transport is nvme, sata,
|
||||
// usb or virtio, and empty when the device path names none, as a RAID
|
||||
// controller's virtual disk does.
|
||||
type DiskFacts struct {
|
||||
Name string `json:"name"`
|
||||
Path string `json:"path"`
|
||||
SizeBytes uint64 `json:"size_bytes"`
|
||||
Model string `json:"model,omitempty"`
|
||||
Transport string `json:"transport,omitempty"`
|
||||
Rotational bool `json:"rotational,omitempty"`
|
||||
Removable bool `json:"removable,omitempty"`
|
||||
}
|
||||
|
||||
// FilesystemFacts is one mounted block-device or ZFS filesystem. BindOf names
|
||||
// the directory a bind mount shows; Disk names the drives under it. Error says
|
||||
// why Usage is missing or stale.
|
||||
type FilesystemFacts struct {
|
||||
Mount string `json:"mount"`
|
||||
Source string `json:"source"`
|
||||
Disk string `json:"disk,omitempty"`
|
||||
Type string `json:"type"`
|
||||
BindOf string `json:"bind_of,omitempty"`
|
||||
Usage *FilesystemUsage `json:"usage,omitempty"`
|
||||
Error string `json:"error,omitempty"`
|
||||
// ID identifies the mount on its node, so a filesystem replaced at the
|
||||
// same path is read afresh. Empty for a mount the daemon cannot see.
|
||||
ID string `json:"-"`
|
||||
}
|
||||
|
||||
// FilesystemUsage is a filesystem's last statfs reading, as df counts it.
|
||||
type FilesystemUsage struct {
|
||||
SizeBytes uint64 `json:"size_bytes"`
|
||||
UsedBytes uint64 `json:"used_bytes"`
|
||||
AvailBytes uint64 `json:"avail_bytes"`
|
||||
}
|
||||
|
||||
// NICFacts is one physical network interface and, where LLDP reports it, the
|
||||
// switch port it is cabled to.
|
||||
type NICFacts struct {
|
||||
Name string `json:"name"`
|
||||
MAC string `json:"mac,omitempty"`
|
||||
Driver string `json:"driver,omitempty"`
|
||||
Firmware string `json:"firmware,omitempty"`
|
||||
BusInfo string `json:"bus_info,omitempty"`
|
||||
Model string `json:"model,omitempty"`
|
||||
SpeedMbps int `json:"speed_mbps,omitempty"`
|
||||
MaxMbps int `json:"max_mbps,omitempty"`
|
||||
Duplex string `json:"duplex,omitempty"`
|
||||
Port string `json:"port,omitempty"`
|
||||
MTU int `json:"mtu,omitempty"`
|
||||
State string `json:"state"`
|
||||
Master string `json:"master,omitempty"`
|
||||
Planes []string `json:"planes,omitempty"`
|
||||
Neighbor *LLDPNeighbor `json:"neighbor,omitempty"`
|
||||
}
|
||||
|
||||
// LLDPNeighbor is what the switch on the other end of a NIC advertises.
|
||||
type LLDPNeighbor struct {
|
||||
ChassisID string `json:"chassis_id"`
|
||||
SystemName string `json:"system_name,omitempty"`
|
||||
SystemDescription string `json:"system_description,omitempty"`
|
||||
PortID string `json:"port_id,omitempty"`
|
||||
PortDescription string `json:"port_description,omitempty"`
|
||||
Capabilities []string `json:"capabilities,omitempty"`
|
||||
}
|
||||
|
||||
// PlaneFacts is the address a plane uses on this node and the NICs under it.
|
||||
type PlaneFacts struct {
|
||||
Plane string `json:"plane"`
|
||||
Bridge string `json:"bridge,omitempty"`
|
||||
Address string `json:"address,omitempty"`
|
||||
Prefix int `json:"prefix,omitempty"`
|
||||
NICs []string `json:"nics,omitempty"`
|
||||
MTU int `json:"mtu,omitempty"`
|
||||
}
|
||||
|
||||
// NetTestServeRequest starts a server bound to the node's address on Plane.
|
||||
type NetTestServeRequest struct {
|
||||
Plane string `json:"plane"`
|
||||
DurationSeconds int `json:"duration_seconds"`
|
||||
}
|
||||
|
||||
// NetTestServeResponse names the address the server listens on.
|
||||
type NetTestServeResponse struct {
|
||||
Address string `json:"address,omitempty"`
|
||||
Port int `json:"port,omitempty"`
|
||||
Error string `json:"error,omitempty"`
|
||||
}
|
||||
|
||||
// NetTestRunRequest runs the client towards Address from the node's own
|
||||
// address on Plane. Address must be what Target reports for that plane.
|
||||
type NetTestRunRequest struct {
|
||||
Plane string `json:"plane"`
|
||||
Target string `json:"target"`
|
||||
Address string `json:"address"`
|
||||
Port int `json:"port"`
|
||||
DurationSeconds int `json:"duration_seconds"`
|
||||
Streams int `json:"streams"`
|
||||
}
|
||||
|
||||
// NetTestRunResponse is one direction of a pair's measurement.
|
||||
type NetTestRunResponse struct {
|
||||
SourceAddress string `json:"source_address,omitempty"`
|
||||
BitsPerSecond float64 `json:"bits_per_second"`
|
||||
Bytes int64 `json:"bytes"`
|
||||
Seconds float64 `json:"seconds"`
|
||||
Streams int `json:"streams"`
|
||||
StreamBPS []float64 `json:"stream_bps,omitempty"`
|
||||
RTTMinMillis float64 `json:"rtt_min_ms,omitempty"`
|
||||
RTTAvgMillis float64 `json:"rtt_avg_ms,omitempty"`
|
||||
RTTMaxMillis float64 `json:"rtt_max_ms,omitempty"`
|
||||
PacketLoss float64 `json:"packet_loss_percent,omitempty"`
|
||||
LinkMbps int `json:"link_mbps,omitempty"`
|
||||
MTU int `json:"mtu,omitempty"`
|
||||
Tool string `json:"tool,omitempty"`
|
||||
Error string `json:"error,omitempty"`
|
||||
}
|
||||
@@ -1,29 +0,0 @@
|
||||
package clusterv1
|
||||
|
||||
import "time"
|
||||
|
||||
// ClusterStateBucket is the KV bucket holding each daemon's heartbeat alongside
|
||||
// other cluster-wide state.
|
||||
const ClusterStateBucket = "spinifex-cluster-state"
|
||||
|
||||
// HeartbeatKey is the ClusterStateBucket key a node writes its heartbeat under.
|
||||
func HeartbeatKey(node string) string {
|
||||
return "heartbeat." + node
|
||||
}
|
||||
|
||||
// HeartbeatTimestamp is the read side of a heartbeat for node liveness. It
|
||||
// decodes only the timestamp, so a reader does not depend on the capacity
|
||||
// figures the daemon writes beside it.
|
||||
type HeartbeatTimestamp struct {
|
||||
Timestamp string `json:"timestamp"`
|
||||
}
|
||||
|
||||
// Time returns when the heartbeat was written. A timestamp that is missing or
|
||||
// unparseable vouches for nothing, so ok is false.
|
||||
func (h HeartbeatTimestamp) Time() (written time.Time, ok bool) {
|
||||
parsed, err := time.Parse(time.RFC3339, h.Timestamp)
|
||||
if err != nil {
|
||||
return time.Time{}, false
|
||||
}
|
||||
return parsed.UTC(), true
|
||||
}
|
||||
@@ -1,35 +0,0 @@
|
||||
package clusterv1_test
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"testing"
|
||||
"time"
|
||||
|
||||
clusterv1 "github.com/mulgadc/spinifex/contracts/cluster/v1"
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
// The bucket and key are the deployed layout: a rolling upgrade has old daemons
|
||||
// writing and new readers reading, so these strings cannot change.
|
||||
func TestHeartbeatKey_IsTheDeployedLayout(t *testing.T) {
|
||||
assert.Equal(t, "spinifex-cluster-state", clusterv1.ClusterStateBucket)
|
||||
assert.Equal(t, "heartbeat.node-a", clusterv1.HeartbeatKey("node-a"))
|
||||
}
|
||||
|
||||
func TestHeartbeatTimestamp_ReadsAFullDaemonHeartbeat(t *testing.T) {
|
||||
full := `{"node":"node-a","epoch":3,"timestamp":"2026-10-30T01:02:03Z","vm_count":4,"allocated_vcpu":8}`
|
||||
var h clusterv1.HeartbeatTimestamp
|
||||
require.NoError(t, json.Unmarshal([]byte(full), &h))
|
||||
|
||||
written, ok := h.Time()
|
||||
require.True(t, ok)
|
||||
assert.True(t, written.Equal(time.Date(2026, 10, 30, 1, 2, 3, 0, time.UTC)))
|
||||
}
|
||||
|
||||
func TestHeartbeatTimestamp_UndatedBeatVouchesForNothing(t *testing.T) {
|
||||
for _, ts := range []string{"", "yesterday"} {
|
||||
_, ok := clusterv1.HeartbeatTimestamp{Timestamp: ts}.Time()
|
||||
assert.False(t, ok, "timestamp %q", ts)
|
||||
}
|
||||
}
|
||||
@@ -23,7 +23,7 @@ subject shape are wire compatibility. Callers and receivers must use this
|
||||
package rather than rebuild those details. `commands_test.go` preserves the
|
||||
version-1 JSON shape and subject form.
|
||||
|
||||
This package owns shared command payloads. It
|
||||
This package owns shared command payloads and the drain-volume response. It
|
||||
does not define every action's reply body, authorize the request, store EC2
|
||||
state or implement NATS transport. Those remain responsibilities of the EC2
|
||||
domain and its runtime composition.
|
||||
@@ -37,13 +37,6 @@ contract or an approved compatibility plan; it is not a package refactor.
|
||||
Only the subject is owned here today.
|
||||
The request (`account_id`) and reply (`vpc_id`, `error`) bodies are still declared as anonymous structs on each side; moving them into this package is a contract change that needs its own characterization first.
|
||||
|
||||
## Volume held elsewhere
|
||||
|
||||
A start refused because another node holds one of the instance's volumes is an `IncorrectState` error payload on the `ec2.start`, `ec2.start.<node>` and `ec2.cmd.<instance-id>` replies, and the AWS body cannot say why.
|
||||
The owning node adds the `X-Volume-Held-Elsewhere: true` header to that reply, the `ec2.start` forwarder relays it, and `StartInstances` wraps `ErrVolumeHeldElsewhere` around the error it rebuilds, so a caller matches it with `errors.Is` rather than reading the message.
|
||||
The owning node sets the header only from the EBS provider's `volume_held_elsewhere` code, which every provider returns for this refusal; no reader matches the message text.
|
||||
`VolumeHeldElsewhereText` is the wording viperblockd puts in that refusal's message, not a signal.
|
||||
|
||||
## Platform inventory
|
||||
|
||||
`PlatformInventorySubject` (`ec2.v1.PlatformInventory`) is a request/reply route on which a daemon in the `spinifex-workers` queue group reports one account's key pairs, placement groups, launch templates and their `ec2-tags` entries from the EC2 owner stores.
|
||||
|
||||
@@ -12,11 +12,6 @@ const (
|
||||
// this contract family. It is the NATS permission/subscription pattern for
|
||||
// the per-instance command route, rather than the broader descendant match.
|
||||
InstanceCommandSubjectWildcard = InstanceCommandSubjectPrefix + "*"
|
||||
|
||||
// HandledHeader is set to "1" on every reply an owner sends to a
|
||||
// modify-instance-metadata-options or set-disable-api-termination command.
|
||||
// An owner that predates those commands cannot set it.
|
||||
HandledHeader = "Spinifex-Ec2-Cmd-Handled"
|
||||
)
|
||||
|
||||
// InstanceCommandSubject returns the targeted command subject for instanceID.
|
||||
@@ -28,23 +23,19 @@ func InstanceCommandSubject(instanceID string) string {
|
||||
}
|
||||
|
||||
// EC2InstanceCommand is the NATS wire format for EC2 instance commands
|
||||
// (stop, terminate, start, reboot, attach/detach-volume, attach/detach-eni,
|
||||
// IAM profile, spot lineage, tags, monitoring, metadata options and
|
||||
// termination protection).
|
||||
// (stop, terminate, start, attach/detach-volume, attach/detach-eni).
|
||||
type EC2InstanceCommand struct {
|
||||
ID string `json:"id"`
|
||||
Attributes EC2CommandAttributes `json:"attributes"`
|
||||
AttachVolumeData *AttachVolumeData `json:"attach_volume_data,omitempty"`
|
||||
DetachVolumeData *DetachVolumeData `json:"detach_volume_data,omitempty"`
|
||||
DrainVolumeData *DrainVolumeData `json:"drain_volume_data,omitempty"`
|
||||
AttachENIData *AttachENIData `json:"attach_eni_data,omitempty"`
|
||||
DetachENIData *DetachENIData `json:"detach_eni_data,omitempty"`
|
||||
IamProfileAssociationData *IamProfileAssociationData `json:"iam_profile_association_data,omitempty"`
|
||||
SpotLineageData *SpotLineageData `json:"spot_lineage_data,omitempty"`
|
||||
InstanceTagsData *InstanceTagsData `json:"instance_tags_data,omitempty"`
|
||||
InstanceMonitoringData *InstanceMonitoringData `json:"instance_monitoring_data,omitempty"`
|
||||
|
||||
InstanceMetadataOptionsData *InstanceMetadataOptionsData `json:"instance_metadata_options_data,omitempty"`
|
||||
DisableApiTerminationData *DisableApiTerminationData `json:"disable_api_termination_data,omitempty"`
|
||||
}
|
||||
|
||||
// EC2CommandAttributes indicates which action the daemon should perform.
|
||||
@@ -54,6 +45,7 @@ type EC2CommandAttributes struct {
|
||||
StartInstance bool `json:"start_instance"`
|
||||
AttachVolume bool `json:"attach_volume"`
|
||||
DetachVolume bool `json:"detach_volume"`
|
||||
DrainVolume bool `json:"drain_volume,omitempty"`
|
||||
RebootInstance bool `json:"reboot_instance"`
|
||||
AttachENI bool `json:"attach_eni"`
|
||||
DetachENI bool `json:"detach_eni"`
|
||||
@@ -62,9 +54,6 @@ type EC2CommandAttributes struct {
|
||||
SetInstanceTags bool `json:"set_instance_tags,omitempty"`
|
||||
RemoveInstanceTags bool `json:"remove_instance_tags,omitempty"`
|
||||
SetInstanceMonitoring bool `json:"set_instance_monitoring,omitempty"`
|
||||
|
||||
ModifyInstanceMetadataOptions bool `json:"modify_instance_metadata_options,omitempty"`
|
||||
SetDisableApiTermination bool `json:"set_disable_api_termination,omitempty"`
|
||||
}
|
||||
|
||||
// AttachVolumeData carries parameters for an attach-volume command.
|
||||
@@ -80,6 +69,33 @@ type DetachVolumeData struct {
|
||||
Force bool `json:"force,omitempty"`
|
||||
}
|
||||
|
||||
// DrainVolumeData carries parameters for a drain-volume command, which flushes
|
||||
// the volume's in-flight writes to S3 before a snapshot reads them. The command
|
||||
// is addressed to the instance the volume is attached to because the drain
|
||||
// socket only exists on the node hosting it.
|
||||
type DrainVolumeData struct {
|
||||
VolumeID string `json:"volume_id"`
|
||||
}
|
||||
|
||||
const (
|
||||
// DrainVolumeStatusDrained is the ack a node returns once the volume's
|
||||
// in-flight writes have reached S3.
|
||||
DrainVolumeStatusDrained = "drained"
|
||||
|
||||
// DrainVolumeStatusNotRunning is the ack a node returns when it still holds
|
||||
// the instance but the VM is not running, so no process is writing to the
|
||||
// volume and its sealed checkpoint is already current. An attachment record
|
||||
// outlives the writer — stop deliberately leaves boot volumes attached — so
|
||||
// the caller cannot tell this from the record alone.
|
||||
DrainVolumeStatusNotRunning = "not-running"
|
||||
)
|
||||
|
||||
// DrainVolumeResponse is the reply to a drain-volume command.
|
||||
type DrainVolumeResponse struct {
|
||||
VolumeID string `json:"volume_id"`
|
||||
Status string `json:"status"`
|
||||
}
|
||||
|
||||
// AttachENIData carries parameters for an attach-network-interface command.
|
||||
type AttachENIData struct {
|
||||
NetworkInterfaceID string `json:"network_interface_id"`
|
||||
@@ -116,23 +132,6 @@ type InstanceMonitoringData struct {
|
||||
Enabled bool `json:"enabled"`
|
||||
}
|
||||
|
||||
// InstanceMetadataOptionsData carries the requested metadata options for a
|
||||
// modify-instance-metadata-options command. Unset fields are left unchanged.
|
||||
type InstanceMetadataOptionsData struct {
|
||||
HttpTokens string `json:"http_tokens,omitempty"`
|
||||
HttpEndpoint string `json:"http_endpoint,omitempty"`
|
||||
HttpProtocolIpv6 string `json:"http_protocol_ipv6,omitempty"`
|
||||
InstanceMetadataTags string `json:"instance_metadata_tags,omitempty"`
|
||||
HttpPutResponseHopLimit *int64 `json:"http_put_response_hop_limit,omitempty"`
|
||||
}
|
||||
|
||||
// DisableApiTerminationData carries the desired termination-protection value
|
||||
// for a set-disable-api-termination command. Carried as data so "false" is an
|
||||
// explicit value, not the absence of one.
|
||||
type DisableApiTerminationData struct {
|
||||
Value bool `json:"value"`
|
||||
}
|
||||
|
||||
// SpotLineageData carries the SIR id stamped onto a spot-launched instance.
|
||||
// Lifecycle is always "spot" for this command, so only the SIR id travels.
|
||||
type SpotLineageData struct {
|
||||
|
||||
@@ -16,6 +16,7 @@ func TestEC2InstanceCommandJSONContract(t *testing.T) {
|
||||
StartInstance: true,
|
||||
AttachVolume: true,
|
||||
DetachVolume: true,
|
||||
DrainVolume: true,
|
||||
RebootInstance: true,
|
||||
AttachENI: true,
|
||||
DetachENI: true,
|
||||
@@ -24,27 +25,16 @@ func TestEC2InstanceCommandJSONContract(t *testing.T) {
|
||||
SetInstanceTags: true,
|
||||
RemoveInstanceTags: true,
|
||||
SetInstanceMonitoring: true,
|
||||
|
||||
ModifyInstanceMetadataOptions: true,
|
||||
SetDisableApiTermination: true,
|
||||
},
|
||||
AttachVolumeData: &AttachVolumeData{VolumeID: "vol-1", Device: "/dev/sdf"},
|
||||
DetachVolumeData: &DetachVolumeData{VolumeID: "vol-2", Device: "/dev/sdg", Force: true},
|
||||
DrainVolumeData: &DrainVolumeData{VolumeID: "vol-3"},
|
||||
AttachENIData: &AttachENIData{NetworkInterfaceID: "eni-1", DeviceIndex: 2},
|
||||
DetachENIData: &DetachENIData{AttachmentID: "eni-attach-1", Force: true},
|
||||
IamProfileAssociationData: &IamProfileAssociationData{InstanceProfileArn: "arn:aws:iam::123456789012:instance-profile/example"},
|
||||
SpotLineageData: &SpotLineageData{SpotInstanceRequestId: "sir-1"},
|
||||
InstanceTagsData: &InstanceTagsData{Tags: map[string]string{"Name": "example"}, TagKeys: []string{"old"}},
|
||||
InstanceMonitoringData: &InstanceMonitoringData{Enabled: true},
|
||||
|
||||
InstanceMetadataOptionsData: &InstanceMetadataOptionsData{
|
||||
HttpTokens: "required",
|
||||
HttpEndpoint: "enabled",
|
||||
HttpProtocolIpv6: "disabled",
|
||||
InstanceMetadataTags: "enabled",
|
||||
HttpPutResponseHopLimit: int64Ptr(3),
|
||||
},
|
||||
DisableApiTerminationData: &DisableApiTerminationData{Value: false},
|
||||
}
|
||||
|
||||
data, err := json.Marshal(command)
|
||||
@@ -57,6 +47,7 @@ func TestEC2InstanceCommandJSONContract(t *testing.T) {
|
||||
"start_instance": true,
|
||||
"attach_volume": true,
|
||||
"detach_volume": true,
|
||||
"drain_volume": true,
|
||||
"reboot_instance": true,
|
||||
"attach_eni": true,
|
||||
"detach_eni": true,
|
||||
@@ -64,73 +55,28 @@ func TestEC2InstanceCommandJSONContract(t *testing.T) {
|
||||
"set_spot_lineage": true,
|
||||
"set_instance_tags": true,
|
||||
"remove_instance_tags": true,
|
||||
"set_instance_monitoring": true,
|
||||
"modify_instance_metadata_options": true,
|
||||
"set_disable_api_termination": true
|
||||
"set_instance_monitoring": true
|
||||
},
|
||||
"attach_volume_data": {"volume_id": "vol-1", "device": "/dev/sdf"},
|
||||
"detach_volume_data": {"volume_id": "vol-2", "device": "/dev/sdg", "force": true},
|
||||
"drain_volume_data": {"volume_id": "vol-3"},
|
||||
"attach_eni_data": {"network_interface_id": "eni-1", "device_index": 2},
|
||||
"detach_eni_data": {"attachment_id": "eni-attach-1", "force": true},
|
||||
"iam_profile_association_data": {"instance_profile_arn": "arn:aws:iam::123456789012:instance-profile/example"},
|
||||
"spot_lineage_data": {"spot_instance_request_id": "sir-1"},
|
||||
"instance_tags_data": {"tags": {"Name": "example"}, "tag_keys": ["old"]},
|
||||
"instance_monitoring_data": {"enabled": true},
|
||||
"instance_metadata_options_data": {
|
||||
"http_tokens": "required",
|
||||
"http_endpoint": "enabled",
|
||||
"http_protocol_ipv6": "disabled",
|
||||
"instance_metadata_tags": "enabled",
|
||||
"http_put_response_hop_limit": 3
|
||||
},
|
||||
"disable_api_termination_data": {"value": false}
|
||||
"instance_monitoring_data": {"enabled": true}
|
||||
}`, string(data))
|
||||
}
|
||||
|
||||
func int64Ptr(v int64) *int64 { return &v }
|
||||
|
||||
// An owner that predates the metadata and termination-protection commands
|
||||
// sends and receives payloads without the new fields.
|
||||
func TestEC2InstanceCommandNewFieldsAreOptionalOnTheWire(t *testing.T) {
|
||||
var n1 EC2InstanceCommand
|
||||
require.NoError(t, json.Unmarshal([]byte(`{
|
||||
"id": "i-0123456789abcdef0",
|
||||
"attributes": {"stop_instance": true},
|
||||
"instance_monitoring_data": {"enabled": true}
|
||||
}`), &n1))
|
||||
require.Equal(t, "i-0123456789abcdef0", n1.ID)
|
||||
require.True(t, n1.Attributes.StopInstance)
|
||||
require.False(t, n1.Attributes.ModifyInstanceMetadataOptions)
|
||||
require.False(t, n1.Attributes.SetDisableApiTermination)
|
||||
require.Nil(t, n1.InstanceMetadataOptionsData)
|
||||
require.Nil(t, n1.DisableApiTerminationData)
|
||||
|
||||
unset, err := json.Marshal(EC2InstanceCommand{ID: "i-1", Attributes: EC2CommandAttributes{StopInstance: true}})
|
||||
func TestDrainVolumeResponseJSONContract(t *testing.T) {
|
||||
data, err := json.Marshal(DrainVolumeResponse{
|
||||
VolumeID: "vol-1",
|
||||
Status: DrainVolumeStatusDrained,
|
||||
})
|
||||
require.NoError(t, err)
|
||||
require.NotContains(t, string(unset), "modify_instance_metadata_options")
|
||||
require.NotContains(t, string(unset), "set_disable_api_termination")
|
||||
require.NotContains(t, string(unset), "instance_metadata_options_data")
|
||||
require.NotContains(t, string(unset), "disable_api_termination_data")
|
||||
}
|
||||
|
||||
func TestInstanceMetadataOptionsDataOmitsUnsetFields(t *testing.T) {
|
||||
data, err := json.Marshal(InstanceMetadataOptionsData{HttpPutResponseHopLimit: int64Ptr(1)})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"http_put_response_hop_limit": 1}`, string(data))
|
||||
|
||||
data, err = json.Marshal(InstanceMetadataOptionsData{})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{}`, string(data))
|
||||
}
|
||||
|
||||
func TestDisableApiTerminationDataKeepsExplicitFalse(t *testing.T) {
|
||||
data, err := json.Marshal(DisableApiTerminationData{Value: false})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"value": false}`, string(data))
|
||||
}
|
||||
|
||||
func TestHandledHeaderName(t *testing.T) {
|
||||
require.Equal(t, "Spinifex-Ec2-Cmd-Handled", HandledHeader)
|
||||
require.JSONEq(t, `{"volume_id":"vol-1","status":"drained"}`, string(data))
|
||||
require.Equal(t, "not-running", DrainVolumeStatusNotRunning)
|
||||
}
|
||||
|
||||
func TestInstanceCommandSubjects(t *testing.T) {
|
||||
|
||||
@@ -1,13 +0,0 @@
|
||||
package ec2v1
|
||||
|
||||
import "errors"
|
||||
|
||||
// VolumeHeldElsewhereText is the message viperblockd gives a lease refusal. It
|
||||
// is wording for the operator; callers classify the refusal by its code.
|
||||
const VolumeHeldElsewhereText = "volume is leased by another owner"
|
||||
|
||||
// ErrVolumeHeldElsewhere marks a start refused because another node still
|
||||
// holds one of the instance's volumes. The owning node sets it from the EBS
|
||||
// provider's held-elsewhere code, and a start reply carries it across NATS as
|
||||
// the X-Volume-Held-Elsewhere header rather than as message text.
|
||||
var ErrVolumeHeldElsewhere = errors.New("a volume is leased by another node")
|
||||
@@ -126,8 +126,8 @@ type Assign struct {
|
||||
ENIPrivateIP string `json:"eniPrivateIp,omitempty"`
|
||||
ENISubnetID string `json:"eniSubnetId,omitempty"`
|
||||
// TaskRoleARN is the task's IAM role (from the taskdef); when set, the agent
|
||||
// serves its credentials at 169.254.170.2 under CredID, a scheduler-minted
|
||||
// secret. An assign without a CredID gets no credentials.
|
||||
// serves its credentials at 169.254.170.2 under CredID, defaulting CredID to
|
||||
// the taskID when the scheduler leaves it empty.
|
||||
CredID string `json:"credId,omitempty"`
|
||||
TaskRoleARN string `json:"taskRoleArn,omitempty"`
|
||||
// ExecutionRoleARN is the task's execution role (from the taskdef); when set,
|
||||
|
||||
@@ -1,76 +0,0 @@
|
||||
# Lambda v1 contracts
|
||||
|
||||
This package will own the Lambda process contracts declared by ADR-0009:S13.
|
||||
It holds the L-GUEST declaration and, in `guest.go`, the connection vocabulary both ends share: the scope, the URI SAN roles, the ALPN identifier and the fw_cfg entry names. The transport and authentication are decided (ADR-0009:S17); the invocation message types and framing are not chosen, so no message types are defined here.
|
||||
It is not an AWS tenant-facing API.
|
||||
|
||||
## L-GUEST: node executor to guest agent
|
||||
|
||||
**Owner:** the Lambda domain.
|
||||
|
||||
**Producers and consumers:** the node-local executor (`runtime/compute/lambda`) and the in-guest Lambda agent (`agents/lambda`), in both directions.
|
||||
|
||||
**Operations:** bootstrap with scoped material, readiness, invoke, completion, bounded logs, shutdown.
|
||||
|
||||
**Transport:** TCP over the microVM's network (ADR-0009:S17).
|
||||
The agent dials one connection to a listener the executor owns for the environment, and the guest exposes no listener for host control.
|
||||
At most one authenticated control connection is active per environment, and the agent may reconnect; how a prior connection and its in-flight work are handled is open (ADR-0009 M1).
|
||||
The network is shared and untrusted: no source address, port or bridge membership is identity or authority.
|
||||
Tenant code may be able to reach the listener, but it cannot open the agent's connection or key material, so any connection it makes fails authentication.
|
||||
|
||||
**Authentication:** mutual TLS 1.3, bound to the environment ID and realization generation, using material the user runtime cannot read (ADR-0009:S17).
|
||||
For each generation the executor creates ephemeral self-signed host and guest certificates whose single URI SAN is `spinifex-lambda://<Region>/<account>/<function ID>/<environment ID>/<generation>/<role>`, with role `host` or `guest`.
|
||||
Each end validates the peer certificate against the one pinned peer certificate and separately requires that URI SAN with the opposite role; hostname verification is not relied on.
|
||||
ALPN carries the L-GUEST major version, `spinifex-lambda-guest/1`, and an unsupported version fails the handshake. Builds with the `lambdatestlink` tag offer only `spinifex-lambda-testlink/3` on both host and guest, whose first frame carries the execution-role credential and function environment, so a test-link end and a release end fail the handshake; release builds are unchanged.
|
||||
In the same builds, the gateway reaches the host daemon on the local abstract socket `spinifex-lambda-testlink-control/4`, and each request carries the runtime-image digest the gateway was built with; the daemon refuses a missing or different digest before starting the attempt, and a gateway and daemon on different socket names never connect.
|
||||
On that socket a request's raw payload is at most 6 MiB, checked by the gateway before encoding and by the daemon after decoding.
|
||||
Its encoded frame is at most the base64 length of 6 MiB plus 64 KiB of metadata headroom and 64 bytes of field slack, and a receiver checks the frame length before allocating; metadata has no separate bound within the frame.
|
||||
The credential and environment travel as their own frame of at most 40 KiB, and the reply has the same payload and frame bounds as the request.
|
||||
The daemon holds at most two connections at once, from accept to close, and leaves the rest unaccepted in the listen backlog.
|
||||
Under TLS 1.3 the guest's handshake can complete before the host refuses its certificate, so the guest treats the connection as accepted only after the host's first message.
|
||||
In the same builds, the function owner moves a created function with stored code from `Pending` to `Active` once this node's daemon serves the pinned image and the cluster configuration has exactly one node; a code copy over the test-link bound stays `Pending`, and `StateReason` names the unmet condition.
|
||||
Release builds never leave `Pending`.
|
||||
The Mulga deploy role installs a test-link build with `deploy_lambda_testlink=true`; this profile does not meet Q-199 fail-closed revocation, and its guests take management addresses whose record can expire (mulga-0hsub).
|
||||
|
||||
**Bootstrap material:** the executor passes these fw_cfg entries under `opt/spinifex/lambda/`, and the agent reads them as root and unloads `qemu_fw_cfg` before user code starts.
|
||||
|
||||
| Entry | Content |
|
||||
|---|---|
|
||||
| `guest-key` | Guest private key, PEM |
|
||||
| `guest-cert` | Guest certificate, PEM |
|
||||
| `host-cert` | The one pinned host certificate, PEM |
|
||||
| `expect` | `<Region>/<account>/<function ID>/<environment ID>/<generation>` |
|
||||
| `host-addr` | Host listener as `host:port` |
|
||||
|
||||
The guest's own interface settings come from `opt/spinifex/netcfg`, the entry other system VMs use.
|
||||
The agent unloads `qemu_fw_cfg` even when an entry is missing, and refuses to start if the fw_cfg tree is still readable afterwards.
|
||||
On a VMM without fw_cfg (the test-link Firecracker executor), the executor attaches the same entries as a tar archive on the second, writable virtio-blk disk, laid out as `by_name/<entry>/raw`.
|
||||
The runtime image's init extracts it to a root-only tmpfs at `/spinifex-cfg`, and the agent reads the entries there, then overwrites the whole disk with zeros and removes the tree; it refuses to start if the tree is still present.
|
||||
Until the message format exists, the agent authenticates and holds the connection but accepts no host message and starts no user code.
|
||||
|
||||
**Authority:** one environment's attempt only.
|
||||
The user runtime never holds these keys, and sharing the guest with the agent gives user code no platform authority.
|
||||
|
||||
**Idempotency and fencing:** an invoke carries the attempt-grant identity and the attempt ID.
|
||||
The agent accepts a run command only under a current grant, and a completion under a superseded grant is discarded and reported.
|
||||
How a grant is enforced and proven ended is open (ADR-0009 M1).
|
||||
|
||||
**Errors and versions:** the agent and executor fail closed on an unknown protocol version or an identity mismatch.
|
||||
|
||||
**Readiness:** environment readiness needs authenticated, current-generation agent evidence; a running hypervisor process does not count.
|
||||
|
||||
**Payloads:** the contract must carry a synchronous payload and response of up to 6 MB and an asynchronous payload of up to 1 MB, plus framing, over a path whose size bound is declared.
|
||||
How payloads move is open (ADR-0009 M3).
|
||||
|
||||
### Compatibility boundary
|
||||
|
||||
The agent ships inside the runtime image pinned by each published version, so it is outside the coordinated release train.
|
||||
The executor supports every L-GUEST major carried by a still-supported runtime image.
|
||||
Retiring an image follows a published runtime-deprecation lifecycle that blocks create and update before it blocks invoke.
|
||||
|
||||
Evidence required before the first agent image ships: a function on an old image invoked by a new executor, and the reverse during rollback.
|
||||
|
||||
## L-RUNTIME
|
||||
|
||||
The guest agent serves the AWS Lambda Runtime API, version `2018-06-01`, to the user runtime at `AWS_LAMBDA_RUNTIME_API`.
|
||||
Spinifex does not version this AWS contract, and conformance is measured against AWS.
|
||||
@@ -1,54 +0,0 @@
|
||||
// Package lambdav1 holds the Lambda process-contract vocabulary shared by the
|
||||
// node executor and the guest agent. See README.md for the declarations.
|
||||
package lambdav1
|
||||
|
||||
import (
|
||||
"fmt"
|
||||
"strings"
|
||||
)
|
||||
|
||||
// ALPNv1 names L-GUEST major version 1; a peer offering any other fails the handshake.
|
||||
const ALPNv1 = "spinifex-lambda-guest/1"
|
||||
|
||||
// FwCfgPrefix prefixes every fw_cfg entry the executor passes the guest.
|
||||
const FwCfgPrefix = "opt/spinifex/lambda/"
|
||||
|
||||
// fw_cfg entry names under FwCfgPrefix.
|
||||
const (
|
||||
EntryGuestKey = "guest-key"
|
||||
EntryGuestCert = "guest-cert"
|
||||
EntryHostCert = "host-cert"
|
||||
EntryScope = "expect"
|
||||
EntryHostAddr = "host-addr"
|
||||
)
|
||||
|
||||
// Roles named as the last URI SAN path segment.
|
||||
const (
|
||||
RoleHost = "host"
|
||||
RoleGuest = "guest"
|
||||
)
|
||||
|
||||
const uriScheme = "spinifex-lambda"
|
||||
|
||||
// Scope names one environment generation:
|
||||
// Region/account/function ID/environment ID/generation.
|
||||
type Scope string
|
||||
|
||||
// ParseScope accepts exactly five non-empty, slash-separated segments.
|
||||
func ParseScope(s string) (Scope, error) {
|
||||
parts := strings.Split(s, "/")
|
||||
if len(parts) != 5 {
|
||||
return "", fmt.Errorf("lambdav1: scope %q: want 5 segments, got %d", s, len(parts))
|
||||
}
|
||||
for _, p := range parts {
|
||||
if p == "" || strings.ContainsAny(p, " \t\r\n?#%") {
|
||||
return "", fmt.Errorf("lambdav1: scope %q: empty or invalid segment", s)
|
||||
}
|
||||
}
|
||||
return Scope(s), nil
|
||||
}
|
||||
|
||||
// URI is the single URI SAN a certificate for role in this scope carries.
|
||||
func (s Scope) URI(role string) string {
|
||||
return uriScheme + "://" + string(s) + "/" + role
|
||||
}
|
||||
@@ -1,20 +0,0 @@
|
||||
package lambdav1
|
||||
|
||||
import (
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
func TestParseScope(t *testing.T) {
|
||||
s, err := ParseScope("ap-southeast-2/000000000001/fn-1/env-1/gen-1")
|
||||
require.NoError(t, err)
|
||||
assert.Equal(t, "spinifex-lambda://ap-southeast-2/000000000001/fn-1/env-1/gen-1/host", s.URI(RoleHost))
|
||||
for _, bad := range []string{
|
||||
"", "a/b/c/d", "a/b/c/d/e/f", "a//c/d/e", "a/b/c/d/e/", "a/b/c/d/e?x", "a/b/c/d/e#x", "a b/b/c/d/e",
|
||||
} {
|
||||
_, err := ParseScope(bad)
|
||||
assert.Error(t, err, bad)
|
||||
}
|
||||
}
|
||||
@@ -8,8 +8,3 @@ administration. It is separate from AWS tenant-facing service contracts.
|
||||
roles and Reed-Solomon configuration. It is an operational status contract,
|
||||
not a generic provider abstraction; do not change its subject or JSON shape
|
||||
incompatibly. Add a new version when that is required.
|
||||
|
||||
`admin.go` holds the `/admin/<Method>` request and response bodies the private admin surface shares with the CLI.
|
||||
`AdminMethodNames` is the list of callable methods, and the gateway router builds its accept set from it.
|
||||
`AccountQuotaRequest`, `AccountQuotaResponse` and `DeleteAccountResponse` stay in the gateway until the `/admin` handlers move, because they embed types the contracts layer may not import.
|
||||
Do not change a field name, JSON tag or omitempty incompatibly; add a new version when that is required.
|
||||
|
||||
@@ -1,210 +0,0 @@
|
||||
package operatorv1
|
||||
|
||||
import "slices"
|
||||
|
||||
// adminMethodNames is the callable /admin/<Method> list. The gateway router
|
||||
// builds its accept set from it, so a name cannot be served without being listed.
|
||||
var adminMethodNames = []string{
|
||||
"CreateAccount",
|
||||
"DeleteAccount",
|
||||
"DescribeAccountDeletion",
|
||||
"GetAccountQuota",
|
||||
"ListAccounts",
|
||||
"PutAccountQuota",
|
||||
}
|
||||
|
||||
// AdminMethodNames returns the callable /admin/<Method> names in sorted order.
|
||||
// Tooling that mints a credential grants these by name, so the list has to be
|
||||
// the one the router serves rather than a copy that can drift out of step.
|
||||
func AdminMethodNames() []string {
|
||||
return slices.Sorted(slices.Values(adminMethodNames))
|
||||
}
|
||||
|
||||
// Deletion job states.
|
||||
const (
|
||||
DeletionStateRunning = "RUNNING"
|
||||
DeletionStateCompleted = "COMPLETED"
|
||||
DeletionStateFailed = "FAILED"
|
||||
|
||||
// DeletionStateExercised ends a non-finalizing run: the account is left
|
||||
// TERMINATING and a later DeleteAccount takes the same operation over.
|
||||
DeletionStateExercised = "EXERCISED"
|
||||
)
|
||||
|
||||
// CreateAccountRequest is the /admin/CreateAccount request body.
|
||||
type CreateAccountRequest struct {
|
||||
Name string `json:"name"`
|
||||
ClientToken string `json:"clientToken"`
|
||||
Source string `json:"source,omitempty"`
|
||||
}
|
||||
|
||||
// CreateAccountResponse is the /admin/CreateAccount success body. It is stored
|
||||
// verbatim (with the secret encrypted) so a replay returns an identical answer.
|
||||
type CreateAccountResponse struct {
|
||||
AccountID string `json:"accountId"`
|
||||
AccountName string `json:"accountName"`
|
||||
AdminUser string `json:"adminUser"`
|
||||
AccessKeyID string `json:"accessKeyId"`
|
||||
SecretAccessKey string `json:"secretAccessKey"`
|
||||
DefaultVpcID string `json:"defaultVpcId"`
|
||||
}
|
||||
|
||||
// DeleteAccountRequest is the /admin/DeleteAccount request body.
|
||||
type DeleteAccountRequest struct {
|
||||
AccountID string `json:"accountId"`
|
||||
AccountName string `json:"accountName"`
|
||||
ClientToken string `json:"clientToken"`
|
||||
Force bool `json:"force,omitempty"`
|
||||
DryRun bool `json:"dryRun,omitempty"`
|
||||
|
||||
// NonFinalizing quiesces the account, cleans the quiesced EC2 platform
|
||||
// kinds and stops. No legacy stage runs and the account record stays.
|
||||
NonFinalizing bool `json:"nonFinalizing,omitempty"`
|
||||
}
|
||||
|
||||
// DescribeAccountDeletionRequest is the /admin/DescribeAccountDeletion body.
|
||||
type DescribeAccountDeletionRequest struct {
|
||||
AccountID string `json:"accountId"`
|
||||
}
|
||||
|
||||
// AccountSummary is one row of /admin/ListAccounts. It carries no credential
|
||||
// material: the caller can already create and delete accounts, but that is no
|
||||
// reason for a listing to hand out keys.
|
||||
type AccountSummary struct {
|
||||
AccountID string `json:"accountId"`
|
||||
AccountName string `json:"accountName"`
|
||||
Status string `json:"status"`
|
||||
CreatedAt string `json:"createdAt,omitempty"`
|
||||
}
|
||||
|
||||
// ListAccountsResponse is the /admin/ListAccounts success body.
|
||||
type ListAccountsResponse struct {
|
||||
Accounts []AccountSummary `json:"accounts"`
|
||||
Count int `json:"count"`
|
||||
}
|
||||
|
||||
// IAMQuiescence is the coordinator's recorded view of IAM quiescence for one
|
||||
// fence. Recording only: it admits nothing. QuiescedScope is filled only by a run
|
||||
// that guarantees a final inventory re-read, and holds as of EvaluatedAt.
|
||||
type IAMQuiescence struct {
|
||||
OperationID string `json:"operationId"`
|
||||
Fence uint64 `json:"fence"`
|
||||
Generation uint64 `json:"generation"`
|
||||
QuiescedScope []string `json:"quiescedScope"`
|
||||
Unresolved []UnresolvedScope `json:"unresolved"`
|
||||
Consumers []ConsumerBasis `json:"consumers"`
|
||||
Writers []AccountedWriter `json:"writers"`
|
||||
Inventory []InventoryStatus `json:"inventory"`
|
||||
EvaluatedAt string `json:"evaluatedAt"`
|
||||
}
|
||||
|
||||
// UnresolvedScope names a scope or writer that blocks quiescence, with a fixed
|
||||
// reason. It is never a timeout.
|
||||
type UnresolvedScope struct {
|
||||
Scope string `json:"scope"`
|
||||
Class string `json:"class"`
|
||||
Reason string `json:"reason"`
|
||||
}
|
||||
|
||||
// ConsumerBasis cites an enforcement consumer and the accepted contract that
|
||||
// lets it count as having applied the generation.
|
||||
type ConsumerBasis struct {
|
||||
Consumer string `json:"consumer"`
|
||||
Basis string `json:"basis"`
|
||||
}
|
||||
|
||||
// AccountedWriter is a writer that cannot be fenced and is covered by the
|
||||
// owner's inventory instead. Ref names the tracked work that closes the gap.
|
||||
type AccountedWriter struct {
|
||||
Writer string `json:"writer"`
|
||||
Basis string `json:"basis"`
|
||||
Ref string `json:"ref"`
|
||||
}
|
||||
|
||||
// InventoryStatus is the owner inventory's status for one kind at evaluation.
|
||||
type InventoryStatus struct {
|
||||
Kind string `json:"kind"`
|
||||
Status string `json:"status"`
|
||||
}
|
||||
|
||||
// Classes of an UnresolvedScope.
|
||||
const (
|
||||
UnresolvedClassFence = "fence"
|
||||
UnresolvedClassConsumer = "consumer"
|
||||
UnresolvedClassWriter = "writer"
|
||||
)
|
||||
|
||||
// Fixed reasons of an UnresolvedScope.
|
||||
const (
|
||||
ReasonFenceMismatch = "fence-mismatch"
|
||||
ReasonIAMReadbackFailed = "iam-readback-failed"
|
||||
ReasonNoAcceptedEnforcement = "no-accepted-enforcement-contract"
|
||||
ReasonNoCompleteOwnerInventory = "no-complete-owner-inventory"
|
||||
ReasonParticipantNotDeclared = "participant-not-declared"
|
||||
ReasonOwnerInventoryIncomplete = "owner-inventory-incomplete"
|
||||
ReasonFinalInventoryRereadNotWired = "final-inventory-reread-not-wired"
|
||||
)
|
||||
|
||||
// ExerciseResult is what a non-finalizing run did after IAM quiescence.
|
||||
type ExerciseResult struct {
|
||||
// Skipped is the fixed reason no cleanup ran; Cleanup is then absent.
|
||||
Skipped string `json:"skipped,omitempty"`
|
||||
Cleanup *PlatformCleanup `json:"cleanup,omitempty"`
|
||||
Accounting []KindAccounting `json:"accounting,omitempty"`
|
||||
}
|
||||
|
||||
// PlatformCleanup is the EC2 platform cleanup result. It carries resource IDs,
|
||||
// fixed reasons and counts only, never names, tags, key material or data.
|
||||
type PlatformCleanup struct {
|
||||
Kinds []CleanupKind `json:"kinds"`
|
||||
Items []CleanupItem `json:"items,omitempty"`
|
||||
Counts []CleanupCount `json:"counts,omitempty"`
|
||||
}
|
||||
|
||||
// CleanupKind is one kind's scan status after the cleanup's own re-scan.
|
||||
type CleanupKind struct {
|
||||
Kind string `json:"kind"`
|
||||
Status string `json:"status"`
|
||||
IncompleteReasons []string `json:"incompleteReasons,omitempty"`
|
||||
}
|
||||
|
||||
// CleanupItem is one resource by its ID with what the cleanup did to it.
|
||||
type CleanupItem struct {
|
||||
ResourceID string `json:"resourceId"`
|
||||
Kind string `json:"kind"`
|
||||
Disposition string `json:"disposition"`
|
||||
Reason string `json:"reason,omitempty"`
|
||||
MemberCount int `json:"memberCount,omitempty"`
|
||||
}
|
||||
|
||||
// CleanupCount is the number of records of one class that have no public ID.
|
||||
type CleanupCount struct {
|
||||
Kind string `json:"kind"`
|
||||
Class string `json:"class"`
|
||||
Disposition string `json:"disposition"`
|
||||
Reason string `json:"reason,omitempty"`
|
||||
Count int `json:"count"`
|
||||
}
|
||||
|
||||
// KindAccounting is the final inventory re-read held against the cleanup result
|
||||
// for one kind. Quiescence is not rewritten from it.
|
||||
type KindAccounting struct {
|
||||
Kind string `json:"kind"`
|
||||
Status string `json:"status"`
|
||||
Reason string `json:"reason,omitempty"`
|
||||
Remaining int `json:"remaining"`
|
||||
}
|
||||
|
||||
// Accounting statuses and fixed reasons of a KindAccounting.
|
||||
const (
|
||||
AccountingAccounted = "accounted"
|
||||
AccountingUnaccounted = "unaccounted"
|
||||
|
||||
AccountingReasonRereadIncomplete = "reread-incomplete"
|
||||
AccountingReasonUnnamedRemainder = "unnamed-remainder"
|
||||
)
|
||||
|
||||
// Fixed reasons an exercise ran no cleanup.
|
||||
const (
|
||||
ExerciseSkippedNoQuiescedKind = "no-quiesced-kind"
|
||||
)
|
||||
@@ -1,86 +0,0 @@
|
||||
package operatorv1
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
func goldenJSON(t *testing.T, v any, want string) {
|
||||
t.Helper()
|
||||
data, err := json.Marshal(v)
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, want, string(data))
|
||||
}
|
||||
|
||||
func TestAdminWire_CreateAccountRequest(t *testing.T) {
|
||||
goldenJSON(t, CreateAccountRequest{Name: "acme", ClientToken: "tok-1", Source: "cli"},
|
||||
`{"name":"acme","clientToken":"tok-1","source":"cli"}`)
|
||||
goldenJSON(t, CreateAccountRequest{Name: "acme", ClientToken: "tok-1"},
|
||||
`{"name":"acme","clientToken":"tok-1"}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_CreateAccountResponse(t *testing.T) {
|
||||
goldenJSON(t, CreateAccountResponse{
|
||||
AccountID: "000000000042", AccountName: "acme", AdminUser: "admin",
|
||||
AccessKeyID: "AKIAEXAMPLE", SecretAccessKey: "secret", DefaultVpcID: "vpc-1",
|
||||
}, `{"accountId":"000000000042","accountName":"acme","adminUser":"admin","accessKeyId":"AKIAEXAMPLE","secretAccessKey":"secret","defaultVpcId":"vpc-1"}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_DeleteAccountRequest(t *testing.T) {
|
||||
goldenJSON(t, DeleteAccountRequest{AccountID: "000000000042", AccountName: "acme", ClientToken: "tok-2", Force: true, DryRun: true},
|
||||
`{"accountId":"000000000042","accountName":"acme","clientToken":"tok-2","force":true,"dryRun":true}`)
|
||||
goldenJSON(t, DeleteAccountRequest{AccountID: "000000000042", AccountName: "acme", ClientToken: "tok-2"},
|
||||
`{"accountId":"000000000042","accountName":"acme","clientToken":"tok-2"}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_DescribeAccountDeletionRequest(t *testing.T) {
|
||||
goldenJSON(t, DescribeAccountDeletionRequest{AccountID: "000000000042"}, `{"accountId":"000000000042"}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_DeletionStates(t *testing.T) {
|
||||
require.Equal(t, "RUNNING", DeletionStateRunning)
|
||||
require.Equal(t, "COMPLETED", DeletionStateCompleted)
|
||||
require.Equal(t, "FAILED", DeletionStateFailed)
|
||||
require.Equal(t, "EXERCISED", DeletionStateExercised)
|
||||
}
|
||||
|
||||
func TestAdminWire_NonFinalizingDeleteRequest(t *testing.T) {
|
||||
goldenJSON(t, DeleteAccountRequest{AccountID: "000000000042", AccountName: "acme", ClientToken: "tok-3", NonFinalizing: true},
|
||||
`{"accountId":"000000000042","accountName":"acme","clientToken":"tok-3","nonFinalizing":true}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_ExerciseResult(t *testing.T) {
|
||||
goldenJSON(t, ExerciseResult{
|
||||
Cleanup: &PlatformCleanup{
|
||||
Kinds: []CleanupKind{{Kind: "key-pair", Status: "observed-clean-at-scan-time"}},
|
||||
Items: []CleanupItem{{ResourceID: "pg-1", Kind: "placement-group", Disposition: "blocked", Reason: "instance-members", MemberCount: 1}},
|
||||
Counts: []CleanupCount{{Kind: "key-pair", Class: "public-key-object-without-metadata", Disposition: "deleted", Count: 2}},
|
||||
},
|
||||
Accounting: []KindAccounting{{Kind: "placement-group", Status: AccountingAccounted, Remaining: 1}},
|
||||
}, `{"cleanup":{"kinds":[{"kind":"key-pair","status":"observed-clean-at-scan-time"}],
|
||||
"items":[{"resourceId":"pg-1","kind":"placement-group","disposition":"blocked","reason":"instance-members","memberCount":1}],
|
||||
"counts":[{"kind":"key-pair","class":"public-key-object-without-metadata","disposition":"deleted","count":2}]},
|
||||
"accounting":[{"kind":"placement-group","status":"accounted","remaining":1}]}`)
|
||||
goldenJSON(t, ExerciseResult{Skipped: ExerciseSkippedNoQuiescedKind}, `{"skipped":"no-quiesced-kind"}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_ListAccountsResponse(t *testing.T) {
|
||||
goldenJSON(t, ListAccountsResponse{
|
||||
Accounts: []AccountSummary{
|
||||
{AccountID: "000000000042", AccountName: "acme", Status: "ACTIVE", CreatedAt: "2026-01-02T03:04:05Z"},
|
||||
{AccountID: "000000000043", AccountName: "bolt", Status: "TERMINATING"},
|
||||
},
|
||||
Count: 2,
|
||||
}, `{"accounts":[
|
||||
{"accountId":"000000000042","accountName":"acme","status":"ACTIVE","createdAt":"2026-01-02T03:04:05Z"},
|
||||
{"accountId":"000000000043","accountName":"bolt","status":"TERMINATING"}],"count":2}`)
|
||||
}
|
||||
|
||||
func TestAdminWire_AdminMethodNames(t *testing.T) {
|
||||
require.Equal(t, []string{
|
||||
"CreateAccount", "DeleteAccount", "DescribeAccountDeletion",
|
||||
"GetAccountQuota", "ListAccounts", "PutAccountQuota",
|
||||
}, AdminMethodNames())
|
||||
}
|
||||
@@ -1,71 +0,0 @@
|
||||
package telemetryv1
|
||||
|
||||
import (
|
||||
"regexp"
|
||||
)
|
||||
|
||||
// LiveSamplePeriodMillis is how often a watched instance is sampled for the
|
||||
// console's live view.
|
||||
const LiveSamplePeriodMillis = 500
|
||||
|
||||
// LiveWatchLeaseMillis is how long one watch keeps the live sampler running.
|
||||
// Watchers renew well inside it, so a viewer that goes away costs at most this.
|
||||
const LiveWatchLeaseMillis = 30_000
|
||||
|
||||
// LiveRingSamples bounds the in-memory history a live sampler keeps for a
|
||||
// reconnecting viewer: five minutes at LiveSamplePeriodMillis.
|
||||
const LiveRingSamples = 5 * 60 * 1000 / LiveSamplePeriodMillis
|
||||
|
||||
// Live subjects sit outside metrics.> so nothing that ingests the periodic
|
||||
// CloudWatch batches ever reads a sub-second rate as a period. The account is part
|
||||
// of the subject, so only the poller for that account's instance answers.
|
||||
const (
|
||||
liveWatchSubjectPrefix = "telemetry.watch.ec2."
|
||||
liveSampleSubjectPrefix = "telemetry.live.ec2."
|
||||
)
|
||||
|
||||
var (
|
||||
liveAccountPattern = regexp.MustCompile(`^[0-9]{12}$`)
|
||||
liveInstancePattern = regexp.MustCompile(`^i-[0-9a-f]{8,17}$`)
|
||||
)
|
||||
|
||||
// ValidLiveTarget reports whether accountID and instanceID are safe to place
|
||||
// in a live subject: no NATS token separators or wildcards can get through.
|
||||
func ValidLiveTarget(accountID, instanceID string) bool {
|
||||
return liveAccountPattern.MatchString(accountID) && liveInstancePattern.MatchString(instanceID)
|
||||
}
|
||||
|
||||
// LiveWatchSubject is where a viewer asks the owning node to keep sampling. A
|
||||
// request on it is answered with the ring since LiveWatch.SinceMillis.
|
||||
func LiveWatchSubject(accountID, instanceID string) string {
|
||||
return liveWatchSubjectPrefix + accountID + "." + instanceID
|
||||
}
|
||||
|
||||
// LiveSampleSubject carries one LiveSample per sampling period while watched.
|
||||
func LiveSampleSubject(accountID, instanceID string) string {
|
||||
return liveSampleSubjectPrefix + accountID + "." + instanceID
|
||||
}
|
||||
|
||||
// LiveWatch is the body of a watch. SinceMillis is the newest sample the
|
||||
// viewer already has; zero asks for the whole ring.
|
||||
type LiveWatch struct {
|
||||
SinceMillis int64 `json:"since_ms,omitempty"`
|
||||
}
|
||||
|
||||
// LiveRing answers a watch request with the samples newer than SinceMillis.
|
||||
type LiveRing struct {
|
||||
Samples []LiveSample `json:"samples"`
|
||||
}
|
||||
|
||||
// LiveSample is one live measurement. Every value is a rate over the interval
|
||||
// ending at TimeMillis, measured against the previous sample's clock.
|
||||
type LiveSample struct {
|
||||
TimeMillis int64 `json:"t"`
|
||||
CPUPercent float64 `json:"cpu"`
|
||||
// NetworkIn and NetworkOut are bytes per second from the guest's side.
|
||||
NetworkIn float64 `json:"netIn"`
|
||||
NetworkOut float64 `json:"netOut"`
|
||||
// PacketsIn and PacketsOut are packets per second from the guest's side.
|
||||
PacketsIn float64 `json:"pktIn"`
|
||||
PacketsOut float64 `json:"pktOut"`
|
||||
}
|
||||
@@ -1,42 +0,0 @@
|
||||
package telemetryv1
|
||||
|
||||
import (
|
||||
"encoding/json"
|
||||
"testing"
|
||||
|
||||
"github.com/stretchr/testify/assert"
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
// The console parses these names, so a rename here breaks its live charts.
|
||||
func TestLiveSampleJSONContract(t *testing.T) {
|
||||
data, err := json.Marshal(LiveSample{
|
||||
TimeMillis: 1793450123250, CPUPercent: 12.5,
|
||||
NetworkIn: 48213, NetworkOut: 1290, PacketsIn: 41, PacketsOut: 12,
|
||||
})
|
||||
require.NoError(t, err)
|
||||
require.JSONEq(t, `{"t":1793450123250,"cpu":12.5,"netIn":48213,"netOut":1290,"pktIn":41,"pktOut":12}`, string(data))
|
||||
}
|
||||
|
||||
func TestValidLiveTarget(t *testing.T) {
|
||||
assert.True(t, ValidLiveTarget("123456789012", "i-0123456789abcdef0"))
|
||||
assert.True(t, ValidLiveTarget("123456789012", "i-0abc1234"))
|
||||
|
||||
for _, tc := range []struct{ account, instance string }{
|
||||
{"", "i-0123456789abcdef0"},
|
||||
{"12345678901", "i-0123456789abcdef0"},
|
||||
{"123456789012", ""},
|
||||
{"123456789012", ">"},
|
||||
{"123456789012", "*"},
|
||||
{"123456789012", "i-0123.abcdef"},
|
||||
{"123456789012", "i-0123456789ABCDEF0"},
|
||||
{"123456789012", "vol-0123456789abcdef0"},
|
||||
} {
|
||||
assert.False(t, ValidLiveTarget(tc.account, tc.instance), "%q %q", tc.account, tc.instance)
|
||||
}
|
||||
}
|
||||
|
||||
func TestLiveSubjectsAreOutsideMetrics(t *testing.T) {
|
||||
assert.Equal(t, "telemetry.watch.ec2.123456789012.i-0abc1234", LiveWatchSubject("123456789012", "i-0abc1234"))
|
||||
assert.Equal(t, "telemetry.live.ec2.123456789012.i-0abc1234", LiveSampleSubject("123456789012", "i-0abc1234"))
|
||||
}
|
||||
@@ -1,53 +0,0 @@
|
||||
package telemetryv1
|
||||
|
||||
import "regexp"
|
||||
|
||||
// Node live subjects mirror the instance ones, keyed by node name. Only the
|
||||
// super-admin account's gateway handler subscribes to them.
|
||||
const (
|
||||
nodeLiveWatchSubjectPrefix = "telemetry.watch.node."
|
||||
nodeLiveSampleSubjectPrefix = "telemetry.live.node."
|
||||
)
|
||||
|
||||
var liveNodePattern = regexp.MustCompile(`^[A-Za-z0-9_-]{1,64}$`)
|
||||
|
||||
// ValidLiveNode reports whether a node name is safe to place in a subject.
|
||||
func ValidLiveNode(node string) bool {
|
||||
return liveNodePattern.MatchString(node)
|
||||
}
|
||||
|
||||
// NodeLiveWatchSubject is where a viewer asks a node to keep sampling itself.
|
||||
// A request on it is answered with a NodeLiveRing.
|
||||
func NodeLiveWatchSubject(node string) string {
|
||||
return nodeLiveWatchSubjectPrefix + node
|
||||
}
|
||||
|
||||
// NodeLiveSampleSubject carries one NodeLiveSample per period while watched.
|
||||
func NodeLiveSampleSubject(node string) string {
|
||||
return nodeLiveSampleSubjectPrefix + node
|
||||
}
|
||||
|
||||
// NodeLiveRing answers a watch with the samples newer than its SinceMillis.
|
||||
type NodeLiveRing struct {
|
||||
Samples []NodeLiveSample `json:"samples"`
|
||||
}
|
||||
|
||||
// NodeLiveSample is one host measurement. Rates are per second over the
|
||||
// interval ending at TimeMillis.
|
||||
type NodeLiveSample struct {
|
||||
TimeMillis int64 `json:"t"`
|
||||
CPUPercent float64 `json:"cpu"`
|
||||
// MemoryUsedPercent is used memory less page cache, as free(1) reports it.
|
||||
MemoryUsedPercent float64 `json:"mem"`
|
||||
Load1 float64 `json:"load1"`
|
||||
// Planes holds the traffic of each plane's physical NICs, keyed wan, lan, vpc.
|
||||
Planes map[string]NodeLivePlane `json:"planes"`
|
||||
}
|
||||
|
||||
// NodeLivePlane is one plane's traffic from the host's side.
|
||||
type NodeLivePlane struct {
|
||||
NetworkIn float64 `json:"netIn"`
|
||||
NetworkOut float64 `json:"netOut"`
|
||||
PacketsIn float64 `json:"pktIn"`
|
||||
PacketsOut float64 `json:"pktOut"`
|
||||
}
|
||||
@@ -51,6 +51,15 @@ request/reply replies.
|
||||
removed its local state and export. These routes and JSON shapes are deployed
|
||||
legacy compatibility behaviour, not a generic EBS-provider interface.
|
||||
|
||||
## Fence route
|
||||
|
||||
`VolumeFencedSubject(node)` is `ebs.<node>.fenced`, or `ebs.fenced` with no
|
||||
node for a single-node daemon. The Viperblock adapter publishes
|
||||
`VolumeFencedEvent` on it after tearing down an export whose volume lease
|
||||
moved to another node; the local daemon subscribes to stop the guest that
|
||||
was using it. Node-addressed, like mount and unmount: the guest is local, and
|
||||
no other node's daemon has anything to stop.
|
||||
|
||||
## Abandon route
|
||||
|
||||
`VolumeAbandonSubject(node)` is `ebs.<node>.abandon`. The daemon uses
|
||||
|
||||
@@ -9,8 +9,9 @@ import (
|
||||
"github.com/stretchr/testify/require"
|
||||
)
|
||||
|
||||
// TestVolumeAbandonSubject_BothForms pins the abandon route: node-addressed,
|
||||
// with an empty-node fallback for a single-node daemon.
|
||||
// TestVolumeAbandonSubject_BothForms pins the abandon route the same way the
|
||||
// fenced route is pinned: node-addressed, with an empty-node fallback for a
|
||||
// single-node daemon.
|
||||
func TestVolumeAbandonSubject_BothForms(t *testing.T) {
|
||||
assert.Equal(t, "ebs.node-a.abandon", viperblocklegacyv1.VolumeAbandonSubject("node-a"))
|
||||
assert.Equal(t, "ebs.abandon", viperblocklegacyv1.VolumeAbandonSubject(""))
|
||||
|
||||
@@ -0,0 +1,21 @@
|
||||
package viperblocklegacyv1
|
||||
|
||||
// VolumeFencedSubject is where a fenced volume is announced so the daemon on
|
||||
// this node can stop the guest that was using it. Node-addressed: the guest is
|
||||
// local, and every other node's daemon has nothing to do about it.
|
||||
func VolumeFencedSubject(node string) string {
|
||||
if node == "" {
|
||||
return "ebs.fenced"
|
||||
}
|
||||
return "ebs." + node + ".fenced"
|
||||
}
|
||||
|
||||
// VolumeFencedEvent announces that a node tore down a volume's export because
|
||||
// the lease moved. Winner is who holds it now, for the operator reading the
|
||||
// guest's stop reason.
|
||||
type VolumeFencedEvent struct {
|
||||
Volume string `json:"volume"`
|
||||
Node string `json:"node"`
|
||||
Winner string `json:"winner"`
|
||||
Reason string `json:"reason"`
|
||||
}
|
||||
Some files were not shown because too many files have changed in this diff Show More
Reference in New Issue
Block a user