Treat monitor-only as terminal before quiet-hours replay and flapping, and reject scheduled or queued escalations when the current alert is monitor-only. Align firing, recovery and delivery diagnosis with that policy; cover ordinary and critical-repeat scheduling plus a policy change after callback queueing.
Change-source: pulse-maintainer
Keep resource maintenance, mute and expected-offline policy terminal in the internal alert callback path. Quiet-hours deferral remains replayable, while delivery diagnosis and event history report the actual operator suppression. Cover pre-reconciliation active alerts and unaffected performance signals.
Change-source: pulse-maintainer
Avoid configuration-history rows when source merges only reorder identity addresses or represent an absent list as empty. Real identity changes remain recorded. The #2319 serial-bearing disk fixture and registry-generation check cover the tags-only and no-op identity paths.
Change-source: pulse-maintainer
Linked Agent SMART disks inherit a PVE instance for presentation. Restrict skipped-poll readback to actual PVE observations so an empty disks/list cannot alternate fabricated inventory, tags and history rows. Pin repeated cycles and the source boundary in the monitoring contracts.
Change-source: pulse-maintainer
Preserve the registry-corroborated PBS-to-Agent link on the service row while selecting the Agent host metrics target in the Backups drawer. Reject PVE-only name collisions and retain the target through a transient missing host row. Recompose the unchanged #1723 runtime repair on current main with fresh desktop and phone browser proof.
Change-source: pulse-maintainer
Relay left public checkout on 2026-09-29: it only ever connected the
Pulse Mobile app, which retires on 31 March 2027, and existing Relay
subscribers now carry Pro entitlements. The app still sold it. Every
Community install saw a "Get alerts on your phone ... Available with
Relay and Pro plans" upgrade panel on Alerts destinations, the plan
screen offered a Relay card with "Remote web access via Relay", and the
settings section was called Remote Access although Relay never reached
the web UI.
The Alerts push panel now renders only on instances that have the relay
feature, with no upsell. The settings section, nav, header and locale
catalogs say Pulse Mobile and carry the retirement date. Community sees
only the Pro comparison card and Relay-tier licenses see none; gated
mobile features name Pro as their minimum plan. The relay feature is
labelled "Pulse Relay (Mobile Connection)" in the catalog, plan copy no
longer claims remote web access, the backend pairing diagnostics point
at Settings > Pulse Mobile, and the user docs, including PRIVACY.md,
state that Relay does not provide remote web UI access.
An existing cpu_thermal chip with an unusable CPU value can still carry valid fan readings. Use a free synthetic CPU-chip key instead of replacing that entry when supplementing Pulse #2311, and cover the collision with a regression test.
Change-source: pulse-maintainer
A named CPU chip with no valid CPU temperature must not suppress an identified SoC thermal reading. Preserve other lm-sensors readings and existing valid CPU precedence.
Change-source: pulse-maintainer
When lm-sensors returns disk or other readings without a CPU chip, retain those readings and add a bounded, identified SoC temperature from sysfs. Preserve existing CPU output and ignore unusable optional sysfs data. Cover combined and precedence paths for Pulse #2311.
Change-source: pulse-maintainer
Fall back to bounded, type-checked thermal-zone and hwmon sysfs readings when lm-sensors is absent or empty. Classify Armada thermal JSON as a CPU reading, preserve nonempty sensor output, and cover units, invalid values, and the agent mapping.
Change-source: pulse-maintainer
Recognise exact node.VMID.name notes on Root PBS snapshots so a newer completed CT backup can replace an older name-only match. Keep colliding VMIDs fail-closed when node and name do not distinguish connections, and cover completed and in-flight attribution for Pulse #2292.
Contract-Neutral: Preserve backup attribution safety without changing API fields, persisted shapes, or source contracts.
Change-source: pulse-maintainer
The unlock screen told Docker users to run `docker exec` from the host,
which does not fit the Unraid, Portainer and TrueNAS app UIs, where the
console already runs inside the container. Those `container_other`
installs are the largest group stalling before first setup: in the week
to 2026-09-28, 158 of 3114 persistent Docker installs never got past
first start, against 94 of 5462 binary and systemd installs.
The startup log also suggested revealing the token "by reading the token
file path". That file holds ciphertext, as the security-privacy contract
requires, so anyone following the hint pasted JSON the unlock screen
rejects. The log now names the command, including the console form, and
the install docs and the contract gain the same guidance.
Contract-Neutral: copy-only: new unlock-screen command row and corrected startup log text; no API payload, lifecycle, storage or primitive change; owning security-privacy clause updated
The agentexec read loop removed a dead session from the agent map but never
closed its done channel, so any dispatch already sent on that session kept
waiting for a result that could never arrive and only returned at its
operation timeout. That's what made the Proxmox cancellation replay test
flaky on the shared CI runner. The restarted action runner dropped its first
session right after registering, the replay had already been sent there,
and the server waited the full 30s while the runner sat reconnected on a
new session.
The read loop now signals the session done on exit, so in-flight typed
dispatches and commands report the disconnect straight away. The
integration test sends the same request again on the runner's next
session when that happens, which is the same durable replay it is there
to prove, still with exactly one mutation. It also keeps the runner's log
and prints it on failure so the cause of any future drop is visible.
Wait for the agent supervisor to check false readiness before asserting its commit gate, and give the race-instrumented fixture a non-expiring window. Hold the monitoring request at target resolution after its history snapshot so replacement is deterministic. Keep production paths and assertions unchanged.
Change-source: pulse-maintainer
Main is failing Backend tests (rest-1) on
TestIssue2285LastUpdateCheckRecordsEffectiveChannelOutcome. The exact
server archive rule (ab0aa1ef0) offers an update only when the release
carries the archive for the host architecture, and the #2285 test
releases carried no assets, so its update-offered cases could never
report available. The two check tests also hard-coded a linux-amd64
archive, which failed on arm64 hosts.
Build every fixture archive from the same runtime asset helper the
updater uses. Test only.
Hourly cleanup previously aged pending image updates by first detection, so a continuing update older than 24 hours lost its tracking and could restart its delay. Track observation activity separately, keep unknown checks pending, and reclaim state when observations stop or an affirmative clear arrives. Cover 24- and 48-hour delays and cleanup.
Contract-Neutral: Preserve the existing Docker update alert policy and public contract while correcting internal tracking expiry.
Change-source: pulse-maintainer
A globally observed second cluster can leave a PBS source apparently mapped to the first while both clusters match a shared-node snapshot. Do not promote a positive but tied guest match into the VMID-only inference path. Regression covers both VM and CT completed and in-flight snapshots with the second cluster visible on another PBS instance.
Contract-Neutral: Preserve existing PBS backup attribution safety without changing API fields, persisted shapes, or source contracts.
Change-source: pulse-maintainer
A newer snapshot from another cluster can share the same typed VMID and node namespace. Compare each positive match with competing guests before selecting by recency; a stronger competitor excludes it and a tie needs independent attribution. Cover VM and CT completed and in-flight snapshots while preserving the newer weaker unique match.
Contract-Neutral: Repair existing backup badge attribution and age selection without changing API fields, persisted shapes, or PBS source contracts.
Change-source: pulse-maintainer
A stronger namespace or comment score currently makes an old snapshot beat a newer, still attributable snapshot. Keep the existing typed-VMID and cross-instance eligibility guards, then choose the latest eligible backup for LastBackup and in-flight state. Add VM and CT regressions for unique fallback and colliding IDs.
Contract-Neutral: Repair existing backup-age selection without changing API fields, persisted shapes, or PBS attribution eligibility.
Change-source: pulse-maintainer
## What changed and why
Keep queued notifications from being cancelled while saved destinations
load, and make stable-channel update discovery work with GitHub's large,
prerelease-heavy release list (#2282). Update download selection now
requires the exact server archive for the runtime architecture; ARMv6
and ARMv7 are distinguished by the build variant rather than silently
choosing ARMv7. The batch also retains the reviewed disk-monitoring
round-trip correction.
## Reviewed source
`b1e75fb9e4693c5706b2242760d6277e5291e7f9..18f2dc1f19416bd83ab72e7f82f2287063b0bedc`
(notification test isolation and startup gate, integration of update
metadata, exact-archive selection, ARM variant correction).
## Validation
Provenance and diff checks passed for the complete range. Prior
committed-source offline proofs passed for the update package through
`ab0aa1ef02698f7727b134a6f6b5ed520930cc61` and for the notification
startup race scenario. The new ARM mapping has a focused regression test
in this head; exact-head protected CI remains required before landing.
These source checks do not establish installed update or alert-delivery
results, nor release readiness.
[How Pulse handles
triage](https://github.com/rcourtman/Pulse/blob/main/docs/AI_TRANSPARENCY.md)
Combine the protected schema-v18 sender with the exact-archive updater. Require installable archives in positive fixtures, classify missing archives and unavailable Pro checks without claiming an offer, and avoid caching an unactivated Pro result. Record the contract and exact combined-source browser receipt.
Change-source: pulse-maintainer
Combine the reviewed main-line sender with the in-flight archive correction so the interaction can be repaired and proved at exact source.
Change-source: pulse-maintainer
A failed update check never reaches the update history, because nothing
was applied, so telemetry could not tell an install whose check is broken
from one that was offered an update and ignored it. The 6.4.3-rc.1 installs
looked healthy while their check failed on every call (#2282).
The update manager now records a closed outcome for every check on the
install's effective channel (up_to_date, available, no_release,
rate_limited, network_error, metadata_error, skipped, error), tagging
lookup failures without changing their error text, and ignores previews of
another channel from Settings. Telemetry schema v18 exports update_channel,
update_check_outcome and update_available. No version, URL or error text
leaves the install. The Settings preview type, privacy disclosure, adoption
report and subsystem contracts are updated to match; the Pulse Pro receiver
change lands separately and the parity check passes against it.
ARMv6 and ARMv7 server binaries both report GOARCH=arm. Read the release binary GOARM build setting and fail closed when it does not name a published server archive, rather than offering an ARMv7 update to ARMv6 installations. Cover the architecture mapping and document the installability boundary.
Change-source: pulse-maintainer
Release metadata contains agent, MCP and multiple architecture archives. An incomplete release must not offer an unrelated tarball or advertise an update that cannot be applied. Retain server archive names only, require the exact tag and runtime architecture, and record this installability boundary with regression coverage.
Change-source: pulse-maintainer
GitHub's release list is now ~9.6 MB because every 6.4.x release embeds
349 assets (~600 KB of JSON each). The update check buffered the whole
response behind a 1 MiB bound, so the primary path always failed and fell
through to the Atom feed. That feed holds ten entries and each release now
spends three of them, so stable-channel checks on main failed outright
("no suitable release found for channel stable") and 6.4.5 stable would
have shipped relying on it.
Decode the release list and the single-release notes response as a token
stream, keeping only tag, name, body, draft, prerelease, publish time and
the Pulse linux server archives. Memory now tracks what the updater keeps,
so the byte bound rises to 64 MiB while staying typed; the Atom feed stays
as the last resort for that over-limit condition only.
Verified against live GitHub: stable resolves v6.4.1 and rc resolves
v6.4.5-rc.3 through the primary path with no feed fallback.
The monitor restored notification destinations after constructing a queue whose five-second ticker could consume due jobs. Defer processor installation through destination restore, then wake the backlog before external alert callbacks admit first agent reports. Cover both forced pre-configuration processing and the actual monitor restart path; update the notifications, monitoring and agent-lifecycle contracts.
Change-source: pulse-maintainer
SetProcessor wakes the live queue worker while this test processes a fetched row manually. Defer the row beyond GetPending before installing the processor so a competing worker cannot decide the synchronous assertion.
Change-source: pulse-maintainer
Keep the persisted empty-device alert ID distinct from the root-device ID while excluding incomplete disks from alias claims. Extend the unified-resources contract and identity tests in the same commit.
Change-source: pulse-maintainer
Keep persisted PVE disk alert identities stable while indexing them as exact physical-disk aliases. A muted disk now clears and suppresses its wearout alert without suppressing neighbouring disks. Update both subsystem contracts and add alert-identity, canonical-alias and registry-backed policy proofs.
Change-source: pulse-maintainer
A single healthy-looking disk poll could resolve a worn SSD alert, so the next
low reading re-fired its occurrence and notification even after the host-agent
SMART merge correction. Feed reported wearout through the canonical reducer
with three recovery confirmations, retain the failing value during a pending
recovery, and leave absent evidence unchanged. Repeated alternating polls now
keep one active occurrence and one dispatch; sustained recovery still clears it.
Refs #2112
Change-source: pulse-maintainer
Carry changed resource IDs into the Workloads projection so periodic deltas
reuse unchanged rows. Index canonical aliases after ingest and resolve
identity-only links without cloning resources. Remove repeated group payloads
from windowed Storage rows, and keep fleet notices and Agent Doctor usable at
narrow widths.
Record the browser proof and remaining payload and benchmark observations in
the owning subsystem contracts.
Refs #2199
Suppressed alert writes previously returned without insertion while
callers continued adding history, recording firing events and requesting
notification delivery. Make acceptance explicit across detector and
restore paths. Repeated muted, retired and maintenance observations now
stop before those side effects, while expiry and resumed monitoring
admit the condition once.
Startup now reconciles restored alerts as soon as persisted resource
policy becomes available, refreshes shared alert state and persists that
resolution. Reconciliation also runs after resource publication, when
restored native aliases first resolve to their canonical agent policy.
Previously muted or retired resources could retain alerts restored
before either boundary.
Retained alert occurrences now reopen when the reducer explicitly
re-fires. Canonical history records occurrence identity separately from
transition time, preserving one incident after checkpoint loss. Delayed
initial firing, resolution replay and historical read repair cannot
override a newer transition. Older canonical records without the
occurrence key retain their existing matching behavior.
Shared table and card renderers preserve components retained across
scrolling windows. Previously a window shift rebuilt those items,
resetting the node Manage tab and losing unsaved lifecycle edits while
scrolling to Save. Mounted regressions cover retained input identity,
text, global position and the existing row budget. Maintenance banners
now show actual local start and end times instead of formatting future
timestamps as “just now”, and describe the paused attention correctly.
The synthetic host fixtures now keep stable machine identities across
graph rebuilds, so saved policies remain attached through restart
verification. The existing stability regression now checks machine IDs
per host, Docker agent IDs and unified canonical agent IDs. Production
conflict checks for a genuinely changed machine identity are unchanged.
Generated installation and rollback guidance now identifies the
Pulse-owned update helper and directs other installations to the
versioned signed installer flow. Both signed installer examples stop on
download or signature failure and run in a fresh temporary directory.
Validation:
- Suppression regression reproduced on the original source across 15
cases, then passed with the change. Full alert package tests pass.
Extended expiry, reconciliation and restore cases pass with the race
detector.
- Signed installer recipes fail six baseline cases and pass all eight
corrected cases. Seven existing installer pinning/signature/rollback
tests pass.
- Release renderer (39 tests), promotion resolver (47 tests),
release-promotion policy (57 tests), documentation mirrors and links,
and governance audits pass. Installer policy assertions cover helper
ownership, the signed fallback, exact version pinning, rollback
applicability and post-restart version verification. The complete
repository pre-commit hook passes on the worker against the matching
staged source, including repository tooling tests, release-control
helper tests and frontend lint. The global sensitivity check passes
locally.
- Startup restore and late native-alias resolution both fail on their
baselines. Final focused proof, race proof and the full monitoring
package pass.
- Retained-occurrence refire regression reproduced in both incident
projections. The complete memory suite passes on final source, full
monitoring and resource-store suites pass, and final focused
replay/refire proof passes under the race detector.
- Shared window regression failed for both table and card renderers on
the baseline, then all nine tests passed. Related component checks
passed 88 tests across seven files. Typecheck and canonical platform
audit pass.
- Maintenance rendering reproduces both incorrect future-time cases on
the baseline. Final operator-state tests pass 22 cases, and targeted
lint and typecheck pass.
- The fixture identity regression fails on the committed generator and
passes after deterministic IDs, together with four related host-link
tests and the Docker host fixture test.
- Production frontend and backend builds pass. Playwright inspection
covers Proxmox Manage and Alerts at 390x844 and 1440x900, 900px History
controls, and shipped install/upgrade docs. Saved maintenance and mute
survive a graceful same-data restart with stable canonical identity.
Cancelling maintenance preserves mute, resuming monitoring reopens the
retained incident with ordered fire/resolve/refire events.
- Authentication and credential baseline passes, including all 41
required API security regressions. All 11 selected Linux peer-identity,
command approval and observer-isolation regressions pass.
- The existing WebSocket connection-cap test now waits for actual per-IP
slot release before testing slot reuse. It still requires HTTP 429 at
capacity and successful registration after release. The corrected test
passes 100 runs under the race detector, and the full agent-execution
package passes with the race detector on the CI merge source.
The existing resource-payload performance gap remains open. The initial
50-node measurement recorded a 1,037 ms maximum desktop idle long task.
A final-source run still observed multi-second tasks, but its elevated
host load prevents a controlled comparison. Both observations and their
limits are retained. This change does not claim to resolve cross-client
payload or browser responsiveness work.
Refs #2237
Agent removal occurs inside read-loop cleanup before the handler releases
its per-IP connection slot. The flood test could observe the first state
and retry while the slot was still occupied, causing a spurious rejection.
Wait for the guarded admission count to reach zero before reconnecting,
while preserving rejection at capacity and successful registration after
release.
Rejected active-alert writes still reached history, firing events and
notification dispatch. Make admission explicit across detector and restore
paths so repeated suppressed observations cannot create those side effects.
Reconcile restored alerts when persisted resource policy attaches at startup,
reconcile again after native aliases become canonical, and refresh shared
state before those alerts reach clients.
Preserve occurrence identity when retained alerts re-fire, including durable
history reconstruction and delayed lifecycle replay.
Keep retained table and card components mounted when scrolling shifts the
visible window, so policy editors do not lose unsaved state. Show actual
future maintenance times and the paused-attention effect.
Keep synthetic host machine identities stable across restarts so saved
operator policy remains attached during verification.
Qualify server update and rollback instructions by helper ownership, and
stop signed installer recipes before execution if a download or signature
check fails.
Refs #2237
The advisory Benchmarks check failed four pull requests on
NormalizeSegment alone in the last 48 paired runs: #2217, #2225, #2252
and #2253. None of them touched the normaliser. On #2252 all five
sub-benchmarks rose 25-39% together. Nine other runs moved the rows
down by more than 10% instead, -20% to -28% on #2228, #2230 and #2242.
Those 13 runs used EPYC 7763, EPYC 9V74 and Intel Xeon runners.
The values are quantized. On EPYC 7763 short_name is 2.50ns or 3.12ns
and uuid is about 29.6ns or 40.5ns, a cycle or two per call, or one
per byte of the UUID scan. Go aligns these functions to 32 bytes, so a
change anywhere earlier in the binary can move them between the halves
of a 64-byte block. A 32-byte shift is enough: on pulse-dev, adding an
unrelated init-time function ahead of the normaliser flipped that
parity and failed the old tree the way #2252 did, all five rows +25% to
+40% at p=0.000 and nothing else flagged. More samples cannot separate
that from a real change, and batching does not help, because every
call repeats the same instructions.
The route-level rows do not behave this way. Over the same 48 runs no
NormalizeRoute row except the root case retired in #2255 moved by more
than 10%. with_uuid came closest at 9.2%.
The guarantees the benchmark carried now live in deterministic tests.
TestNormalizeSegment, TestLooksLikeUUID and TestIsNumeric already pin
classification. TestNormalizeSegment_DoesNotAllocate pins zero
allocations for the five former inputs, and fails with one allocation
per call when the pass-through return copies the segment. It stores
each result in a package-level sink: when normalizeSegment was cut
down to an inlinable copy, a discarded result let the compiler keep
short segments on the stack, and the test missed the allocation.
Timing regressions still show in BenchmarkNormalizeRoute, since every
path runs its segments through normalizeSegment. The proofs below are
paired runs of the CI runner's recipe on pulse-dev, narrowed to
./internal/api/ and raised to 30 samples, because the shared host's
load spread these rows by up to 25%, against about 1% on CI runners.
- looksLikeUUID scanning twice: check-bench-regression.sh exited 1,
flagging only NormalizeRoute/with_uuid, +13.7% (p=0.005).
- normalizeSegment doing its work twice: exited 1, flagging only
NormalizeRoute/with_uuid, +21.9% (p=0.030).
- the candidate against a copy of itself: exited 0.
- the candidate against itself plus the same init-time function:
exited 0. The route rows still moved 4-5%, and with_uuid 9.2%
(p=0.000).
The margin for a UUID-scan slowdown is thin. On EPYC 7763 the scan is
about 30ns of a 280ns with_uuid call, so doubling it reads as roughly
+11% to +14%, while layout alone has moved with_uuid by up to 9.2% both
on CI and in the probe above. Giving UUIDs a larger share of a route
would make that row flip the way this one did; a smaller share would
hide the slowdown.
Preserve the reviewed provider Plan return history and incorporate current upstream changes. Rebuild the portal bundle and manifest from the combined source.
Change-source: pulse-maintainer
The advisory Benchmarks check kept failing unrelated pull requests on
NormalizeRoute/root alone (#2228, #2242, #2245, #2250, and PR 1973 on
2026-09-08), and flagged it beside other results on #2230 and #2244.
The "/" path is a length compare and a constant return.
In the last 19 paired PR runs, the 11 on AMD EPYC 7763 measured root
at exactly 2.18ns or 2.50ns on both sides, each within 0-1%. The two
values are about one CPU cycle apart. In five of those runs base and
candidate landed on different values, four times up (+14%) and once
down (-13%), though none touched the normaliser. The earlier
investigation found the benchmark loop identical and the normaliser
differing only in relocated addresses. More samples cannot separate a
stable one-cycle shift from a real change, and a looser threshold
would blunt the gate for everything else.
Batching root calls does not escape it. On an amd64 host a 64-call
batch spread as widely in relative terms as a single call, because
every call repeats the same few instructions.
What the path must guarantee is that "/" and "" return before
splitting. TestNormalizeRoute_RootFastPathDoesNotAllocate pins that
deterministically: with the fast path removed, TestNormalizeRoute
still passes, but the new test fails with one allocation per call.
Real regressions stay visible. A paired run using the CI runner's own
alternating recipe (10 samples, 100ms), with normalizeRoute made to do
its work twice, flagged every remaining NormalizeRoute sub-benchmark
at +89% to +107%, FullMiddlewarePath at +54%, and bytes and allocs at
+100%. check-bench-regression.sh exited 1. The same tree compared with
itself exited 0.
A provider-hosted control plane refused to start once its licence was
past
expiry and the 7-day grace, because startup validated it with
ValidateLicense. So at the next restart after a 60-day evaluation ran
out,
or after a paying provider's subscription lapsed, the portal went down,
and
the portal's Plan tab is the only place a provider can buy or renew.
Reproduced on the walkthrough lab.
Startup now accepts an authentic, key-bound licence even when lapsed
(ValidateLicenseAllowingLapse), preferring a current licence and
otherwise
the one that expired later, and logs the lapse. It unlocks nothing:
client
runtimes still verify the provider licence in every lease with
ValidateLicense and drop MSP capabilities, new clients are refused with
provider_msp_license_lapsed, and the refresher still refuses a lapsed
licence. The Plan panel says the evaluation or plan ended and offers the
plans, and provider-msp status reports license_lapsed without failing so
a
lapsed install can still upgrade.
The same run showed the portal printing control plane error codes such
as
provider_msp_license_lapsed instead of their message. The portal now
prefers an error's message over its code, which also fixes
already_subscribed and checkout_unavailable in the Plan tab.
Replaces #2241 and #2250, replayed onto current main as one commit.
Every accepted agent report refreshed the canonical resource store, and
every refresh covers the whole estate, so refresh work grew with agent
count × estate size (#2199). #2235 made each pass linear and cheaper.
This PR bounds how many passes reports can trigger.
**Behaviour.** Accepted host, Docker and Kubernetes reports, including
rejected-report liveness, refresh through
`refreshUnifiedResourceStoreAfterAgentReport`:
- The first report after a quiet window is published before its handler
returns, the same as today.
- Reports within 2 s of the previous refresh fold into one trailing
refresh at the window's end. It reads the latest state and broadcasts it
to websocket clients.
- 2 s matches `readPathRegistryFreshness`, the staleness the read paths
already accept.
- Agent removal and host-agent evaluation still refresh immediately, so
removed inventory retires at once.
- `Monitor.Stop` cancels a pending trailing refresh and waits for a
running one before closing stores.
- Monitors built without `New` keep a zero window and refresh on every
report.
**Cost bound.** At 2,080 resources one refresh costs about 82 ms (after
#2235). 200 agents reporting every 30 s drive 6.7 refreshes/s, roughly
55% of a core; coalesced, reports trigger at most one refresh per 2 s,
about 4%, however many agents report.
**Contract change.** The agent-lifecycle contract promised that an
accepted report is visible when its handler returns. It is now visible
within the 2 s window. The founder approved that trade on 24 September
2026, and the contract records it.
**Tests**
- `TestAgentReportRefreshFoldsReportsWithinWindow`: the leading report
is published synchronously; a burst inside the window is deferred; one
trailing refresh publishes the burst's latest value; after a quiet
window the next report is synchronous again. It fails when the throttle
is removed.
- `TestStopCancelsPendingAgentReportRefresh`: fails if stop does not
cancel the pending refresh.
- The issue-1753 end-to-end test posts two reports back to back, so it
now waits for the published rows instead of reading them at once.
**Verification on pulse-dev**
- `internal/monitoring` runs in 112 s against 107 s on #2235's head. No
test is slower by more than 1.2 s.
- `internal/unifiedresources`, `internal/websocket` and
`internal/api/...` pass. `TestServerInfoEndpointReportsDevelopment`
fails in that source export with and without this change, because the
export has no git checkout.
Refs #2199
Follow-up to #2234. Patrol stopped calling
`ChangeDetector.DetectChanges` when it moved to agentic execution
(1b1c9bb2a3). Recent-change context comes from the unified-resource
timeline, and the detector survives only as a fallback reader of the
legacy `ai_changes.json` that older releases wrote. #2234 removed the
detector's unjoined save. The detection code itself was still there, and
only tests used it.
## Removed
- `DetectChanges` and `detectResourceChanges`, which diffed snapshots.
- `ResourceSnapshot` and its `internal/ai` alias.
- Change-ID generation, and the seven description formatters with their
`formatBytes`/`formatFloat` helpers.
- The detector's lock. The history is loaded once at construction and
never modified, so reads need no lock.
- Tests of the detection logic, its heap stability
(`change_detector_memory_regression_test.go`) and ID uniqueness
(`changes_concurrency_test.go`).
## Unchanged
- `NewChangeDetector`, legacy loading and trimming.
- The read APIs: `GetChangesForResource`, `GetRecentChanges` and
`GetChangesSummary`.
- The `ChangeType` constants, which label the legacy entries.
- `intToString`, which the incident and remediation code share.
## Tests
The `internal/ai` tests that used detection to seed the fallback now
write a legacy `ai_changes.json` and load it. That is the path
production reads.
To check that the new seeding really exercises the load path, I broke
legacy loading through an overlay. 10 of the 12 re-seeded tests then
fail. The other two assert that the fallback is left out of a
type-scoped patrol, or that it loses to the canonical timeline, so empty
history also satisfies them.
That run also exposed three weak assertions, now tightened:
- Two formatter comparisons passed against an empty detector because `""
== ""`. They now reject empty output.
- One summary test used `t.Error` and then indexed `[0]`. It now uses
`t.Fatal`.
The ai-runtime contract now describes the detector as a legacy-history
reader.
A provider-hosted control plane refused to start once its licence was past
expiry and the 7-day grace, because startup validated it with
ValidateLicense. So at the next restart after a 60-day evaluation ran out,
or after a paying provider's subscription lapsed, the portal went down, and
the portal's Plan tab is the only place a provider can buy or renew.
Reproduced on the walkthrough lab.
Startup now accepts an authentic, key-bound licence even when lapsed
(ValidateLicenseAllowingLapse), preferring a current licence and otherwise
the one that expired later, and logs the lapse. It unlocks nothing: client
runtimes still verify the provider licence in every lease with
ValidateLicense and drop MSP capabilities, new clients are refused with
provider_msp_license_lapsed, and the refresher still refuses a lapsed
licence. The Plan panel says the evaluation or plan ended and offers the
plans, and provider-msp status reports license_lapsed without failing so a
lapsed install can still upgrade.
The same run showed the portal printing control plane error codes such as
provider_msp_license_lapsed instead of their message. The portal now
prefers an error's message over its code, which also fixes
already_subscribed and checkout_unavailable in the Plan tab.
While a standalone agent stays offline across a restart (up to the
72-hour continuity window), every canonical read-state lookup adds it
back from host continuity by building a whole registry clone
(`readStateWithRecords`). Alert evaluation makes one lookup per resource
per poll, so in that state each poll rebuilt the registry once per
resource. Each lookup also called `GetLiveHostsSnapshot` twice, and that
went through `State.GetSnapshot()`, deep-copying every VM and container
just to return the hosts.
**Changes**
- **Overlay cache.** The adapter keeps the last continuity overlay,
keyed by:
- the registry pointer;
- `LastRebuiltAt` (every replace and in-place supplemental ingest
updates it);
- a fingerprint of the requested records. The fingerprint leaves out
`Resource.UpdatedAt`, which `resourceFromHost` stamps with `time.Now()`
at build time.
It reuses the overlay for at most 2 s, because the overlay evaluates
source staleness when it is built. The live adapter is never mutated.
- **Hosts-only copy.** `GetLiveHostsSnapshot` now copies only hosts,
through the existing `State.GetHosts()`. The result is identical,
because `GetSnapshot` sets `Hosts: cloneHosts(s.Hosts)`. This also
removes a full state copy from each agent report (`ApplyHostReport`),
each config fetch and every other caller.
- **Single host copy per lookup.**
`readStateWithStandaloneHostContinuity` takes the host list once instead
of twice.
**One read-state lookup** with one offline standalone agent (synthetic,
8-node clusters, pulse-dev):
| Resources | main | this PR |
|---|---|---|
| 130 | 3.0 ms / 1.9 MB | 13 µs / 10 KB |
| 520 | 15.9 ms / 11.8 MB | 25 µs / 29 KB |
| 1,040 | 41 ms / 28 MB | 39 µs / 54 KB |
| 2,080 | 136 ms / 95 MB | 74 µs / 104 KB |
**Tests.** Each fails when its mechanism is reverted.
- `TestHostContinuityOverlayReusedWithinRegistryGeneration`: reuse
within a generation, including records rebuilt with a fresh `UpdatedAt`;
rebuild after a new generation or different records; the 2 s age bound;
and no mutation of the live adapter.
- `TestStandaloneHostContinuityReadStateReusedAcrossLookups`: end to end
through `GetUnifiedReadStateOrSnapshot` with a real continuity store.
- `TestGetLiveHostsSnapshotCopiesOnlyHosts`: 1 host with 1,000 VMs must
not allocate per guest. The old path made 1,010 allocations.
**Verification on pulse-dev:** `internal/unifiedresources`,
`internal/monitoring`, the `internal/api/...` subpackages and
`cmd/pulse-control-plane` pass.
`TestServerInfoEndpointReportsDevelopment` fails in that source export
with and without this change, because the export has no git checkout.
Refs #2199
Every accepted agent report sends a websocket state invalidation, and
each state broadcast rebuilds and diffs the full frontend state:
| Resources | One broadcast (`BuildFrontendState` + JSON) |
|---|---|
| 130 | 7.5 ms |
| 1,040 | 62 ms |
| 2,080 | 132 ms |
The hub's keyed snapshot adds about 10 ms per 1,000 resources.
The hub coalesced invalidations with a 100 ms debounce that **restarted
on every signal**. That caused two problems:
- Reports spaced more than 100 ms apart (typical for agents on
independent 30 s timers) each produced a full broadcast. With 200 agents
that is roughly 3.4 broadcasts/s, about half a core whenever a browser
is open.
- Signals closer together than the window postponed the broadcast until
they stopped.
**Change: coalescing is now a throttle.**
- The first invalidation after a quiet period still broadcasts after the
100 ms window.
- Later invalidations for the same audience (all clients, or one tenant)
replace the pending request without rescheduling it.
- They broadcast no sooner than `defaultStateBroadcastInterval` (2 s)
after that audience's previous broadcast. That caps broadcast work near
7% of a core at the same scale.
2 s matches the report-driven store refresh interval (#2242), so faster
broadcasts mostly re-sent the same resources. Broadcasts are still
skipped when no client is connected.
**Tests.** Each fails with the old restart-on-signal, unspaced
behaviour.
- `TestStateBroadcastsSpacedByInterval`: the second broadcast waits for
the interval and carries the latest of a burst; the burst produces only
one broadcast.
- `TestTenantStateBroadcastsSpacedByInterval`: the same for tenant
broadcasts.
- `TestStateBroadcastNotStarvedBySteadySignals`: signals every 10 ms
against a 60 ms window still produce broadcasts. The old debounce
produced none in 600 ms.
**Verification on pulse-dev:** `internal/websocket` (existing coalescing
tests unchanged), `internal/monitoring`, `pkg/server` and the
`internal/api/...` subpackages pass.
`TestServerInfoEndpointReportsDevelopment` fails in that source export
with and without this change, because the export has no git checkout.
Refs #2199
Patrol stopped calling DetectChanges when it moved to agentic execution
(1b1c9bb2a3). Recent-change context comes from the unified-resource
timeline, and the detector survives only as a fallback reader of the
legacy ai_changes.json that older releases wrote. The previous change
removed the detector's unjoined save; the detection itself was still
there, kept alive only by tests.
Remove DetectChanges, its snapshot diffing, ResourceSnapshot (and its
internal/ai alias), change-ID generation, the description formatters and
the detector's lock. The history is now loaded once and never modified,
so reads need no lock. The read APIs, trimming and legacy loading are
unchanged.
Tests that used detection to seed the fallback now write a legacy
ai_changes.json and load it, the same path production reads. With
loading broken, 10 of the 12 re-seeded internal/ai tests fail; the other
two assert that the fallback is absent or loses to the canonical
timeline. Two formatter comparisons now reject empty output, which let
them pass against an empty detector. Tests of the detection logic, its
heap stability and ID uniqueness go with it.
Agents poll `/api/agents/agent/{id}/config` every minute, and every
successful poll was written as a signed `agent_config_fetch` audit
event. On a five-agent homelab install that was 483,000 of 483,101 audit
rows (250 MB) over 75 days, burying the 64 logins and 25 CSRF failures
the log exists to show. At the default 90-day retention it comes to
roughly 60 MB per agent, so an install with hundreds of agents would
carry gigabytes of routine polling in its audit log.
**New rule**
- Every **failed** fetch is still audited (missing scope, unknown agent,
signing or serialisation failure).
- A **successful** fetch is audited only when it is new audit
information:
- the agent's first delivery after Pulse starts;
- a different token;
- a different delivered config (`desiredConfig` hash);
- an unchanged delivery last recorded 24 hours earlier.
- Recorded successes now carry `config=<desiredConfig hash>` and
`reason=first_since_start|token_changed|config_changed|daily` in their
details.
That is about 1,440 rows per agent per day down to a handful.
The tracker lives on the handler, is keyed by org and agent, and is
bounded to 4,096 entries. Handlers built without the constructor keep
the old audit-every-success behaviour.
**Existing rows** are left to age out under retention rather than
deleted, because removing entries from a security log would look like
tampering.
**Docs and contracts.** `AUDIT_LOGGING.md` and its shipped mirror
describe the new rule. The api-contracts and agent-lifecycle contracts
record it. storage-recovery gets a boundary note, because it lists
`internal/api/` as an extension point: recovery actions keep their own
action audits, receipts and verification.
**Tests**
- `TestAgentConfigFetchAuditTrackerRecordsOnlyNewDeliveries` walks first
delivery, unchanged poll, config change, token rotation, and the 24-hour
boundary, plus org isolation and the nil tracker.
- `TestAgentConfigFetchAuditsNewDeliveriesAndEveryFailure` makes five
successful fetches around one config change and gets two success events,
and two failed fetches give two failure events. Without the tracker it
records all five successes.
**Verification on pulse-dev:** `internal/api/...` passes.
`TestServerInfoEndpointReportsDevelopment` fails in that source export
with and without this change, because the export has no git checkout.
pulse-triage[bot]