mirror of
https://github.com/rcourtman/Pulse.git
synced 2026-10-23 12:15:42 +00:00
Prepare v6.5.0-rc.1
Release candidate cut from main at 2e9404504b. It carries every change merged since stable v6.4.5.
Contract-Neutral: Release-only v6.5.0-rc.1 version, notes and packet metadata generated by pulse-maintainer-release-metadata; no public install or runtime contract changes.
Change-source: pulse-maintainer
This commit is contained in:
pulse-triage[bot]
@@ -2,9 +2,9 @@ apiVersion: v2
|
||||
name: pulse
|
||||
description: Helm chart for deploying the Pulse hub and optional Docker, Kubernetes, or OpenShift monitoring agent.
|
||||
type: application
|
||||
version: 6.4.3-rc.1
|
||||
appVersion: "6.4.3-rc.1"
|
||||
icon: https://raw.githubusercontent.com/rcourtman/Pulse/v6.4.3-rc.1/docs/images/pulse-logo.svg
|
||||
version: 6.5.0-rc.1
|
||||
appVersion: "6.5.0-rc.1"
|
||||
icon: https://raw.githubusercontent.com/rcourtman/Pulse/v6.5.0-rc.1/docs/images/pulse-logo.svg
|
||||
keywords:
|
||||
- monitoring
|
||||
- proxmox
|
||||
@@ -32,7 +32,7 @@ annotations:
|
||||
description: Smoke tests with kind cluster deployment
|
||||
artifacthub.io/links: |
|
||||
- name: Documentation
|
||||
url: https://github.com/rcourtman/Pulse/blob/v6.4.3-rc.1/docs/KUBERNETES.md
|
||||
url: https://github.com/rcourtman/Pulse/blob/v6.5.0-rc.1/docs/KUBERNETES.md
|
||||
- name: Support
|
||||
url: https://github.com/rcourtman/Pulse/discussions
|
||||
artifacthub.io/maintainers: |
|
||||
|
||||
@@ -1,6 +1,6 @@
|
||||
# pulse
|
||||
|
||||
  
|
||||
  
|
||||
|
||||
Helm chart for deploying the Pulse hub and optional Docker, Kubernetes, or OpenShift monitoring agent.
|
||||
|
||||
|
||||
+1
-1
@@ -2,7 +2,7 @@ version: '3.8'
|
||||
|
||||
services:
|
||||
pulse:
|
||||
image: ${PULSE_IMAGE:-rcourtman/pulse:6.4.3-rc.1}
|
||||
image: ${PULSE_IMAGE:-rcourtman/pulse:6.5.0-rc.1}
|
||||
container_name: pulse
|
||||
restart: unless-stopped
|
||||
logging:
|
||||
|
||||
@@ -7,8 +7,8 @@ For historical v4 notes that previously lived in this repo, see:
|
||||
`docs/releases/RELEASE_NOTES_v4.md`
|
||||
|
||||
For the current v6 release candidate packet, see:
|
||||
- `docs/releases/RELEASE_NOTES_v6.4.3-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.4.3-rc.1.md`
|
||||
- `docs/releases/RELEASE_NOTES_v6.5.0-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.5.0-rc.1.md`
|
||||
|
||||
For the current stable v6 packet, see:
|
||||
- `docs/releases/RELEASE_NOTES_v6.4.1.md`
|
||||
|
||||
+2
-2
@@ -4,8 +4,8 @@ This guide covers practical upgrade steps for existing Pulse installs moving to
|
||||
|
||||
For the current v6 release candidate packet, see:
|
||||
|
||||
- `docs/releases/RELEASE_NOTES_v6.4.3-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.4.3-rc.1.md`
|
||||
- `docs/releases/RELEASE_NOTES_v6.5.0-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.5.0-rc.1.md`
|
||||
|
||||
For the current stable v6 packet, see:
|
||||
|
||||
|
||||
@@ -0,0 +1,133 @@
|
||||
# Pulse v6.5.0-rc.1 Release Notes
|
||||
|
||||
This release candidate opens the `v6.5.0` candidate line after stable `v6.4.5`, which remains the rollback target.
|
||||
|
||||
Pulse v6.5.0 makes everyday monitoring more dependable, especially when backups are running, connections drop or readings are incomplete. Alerts recover more reliably, backup coverage is clearer, and charts distinguish missing data from real measurements. Patrol gains weekly summaries and clearer cost controls, while setup and updates give you more control over credentials and unattended changes.
|
||||
|
||||
## Highlights
|
||||
|
||||
- Understand what needs attention with clearer readings, backup coverage and delivery status.
|
||||
- Keep monitoring steady through connection failures, restarts and busy systems.
|
||||
- See what Patrol accomplished and make informed choices about its running costs.
|
||||
|
||||
## What's improved
|
||||
|
||||
- **Details stay open and current.** Dashboard refreshes preserve expanded rows, selected tabs, focus and unsaved connection edits without shifting the page.
|
||||
- **History is easier to inspect.** Charts support keyboard and touch inspection, align readings to one time window and show individual observations without inventing a trend.
|
||||
- **Failed refreshes are clearer.** History distinguishes empty results from refresh failures and discards responses for old targets, ranges or organisations.
|
||||
- **Tables are easier to read.** Storage and platform tables fit values more consistently, metric labels use their actual width, and filesystem paths remain readable.
|
||||
- **Navigation is more accessible.** Search history, the command palette, tabs, dialogs and Assistant suggestions work more consistently with keyboards and screen readers.
|
||||
- **Kubernetes job timing is clearer.** Controller details show job times as readable ages, helping you understand when work started or finished.
|
||||
|
||||
## Alerts and notifications
|
||||
|
||||
- **Recurring alerts deliver reliably.** Resolution, restart and delayed retries preserve each occurrence without replaying notifications from an earlier incident.
|
||||
- **Recovery messages match delivery.** Pulse tracks destination results for partial deliveries and distinguishes a notification being dispatched from its receipt.
|
||||
- **Retries handle failures better.** Temporary destination failures respect retry delays, while permanent credential, SMTP and configuration failures stop futile retries.
|
||||
- **Quiet hours remain effective.** Queued notifications are checked against current quiet hours and local time before sending.
|
||||
- **Escalations reach you.** Severity increases can notify through cooldowns, and continuing symptoms can notify after the primary problem recovers.
|
||||
- **Suppression remains respected.** Monitor-only alerts, operator suppression and warning preferences stay effective during escalation and queued delivery.
|
||||
- **Acknowledgements survive cleanup.** Active acknowledgements and observed unacknowledged alerts remain intact, including automatic acknowledgement across repeat firings.
|
||||
- **Flapping follows your settings.** Thresholds above ten are honoured, and expired flapping episodes stop holding later notifications.
|
||||
- **Missing readings do not imply recovery.** Invalid metrics and uncertain connectivity no longer clear storage incidents or produce misleading capacity decisions.
|
||||
- **Delivery status stays current.** Overlapping requests and monitoring reloads no longer replace current notification health with stale results.
|
||||
- **Incident history stays relevant.** Alert timelines and Assistant handoffs retain the correct resource and occurrence, including Docker events.
|
||||
- **Webhook tests match saved settings.** Custom fields use the configured values, and destination server failures are easier to identify.
|
||||
|
||||
## Disks and storage
|
||||
|
||||
- **Disk readings stay tied to their source.** Physical disks retain their identity across refreshes, and SMART history avoids replaying old measurements as new samples.
|
||||
- **Disk health is more accurate.** Endurance values no longer inflate remaining life, and Proxmox disk alerts use the correct physical-disk policy and recovery readings.
|
||||
- **Included mounts remain included.** Explicit disk selections survive ingestion, including tmpfs mounts, and collection uses the agent's mount namespace.
|
||||
- **RAID details are clearer.** VM details show linked disks and RAID, while MD RAID spare counts are corrected without hiding failed members.
|
||||
- **Unraid avoids misleading warnings.** Empty slots and explicitly pool-only arrays no longer trigger inappropriate filesystem or parity warnings.
|
||||
- **Storage details preserve missing data.** Unavailable pool capacity stays unavailable, disk facts remain on their owning host, and stored disk history remains accessible when collection stops.
|
||||
- **Temperatures appear more consistently.** Disk temperature can display without extended SMART data, with improved CPU and other sensor fallbacks, including ARM systems.
|
||||
|
||||
## Proxmox, PBS and backups
|
||||
|
||||
- **Guest reads respect backup locks.** Pulse coordinates guest-agent checks around backups and defers uncertain requests without automatically repeating commands.
|
||||
- **Retained guest readings show their age.** Memory survives repeated backup locks with its original observation time, while paused or incomplete filesystem readings remain clearly unavailable or last known.
|
||||
- **Backup coverage uses the right guest.** PBS snapshots and structured comments are matched only to attributable guests, avoiding shared-node and cross-site mix-ups.
|
||||
- **Backup dates stay honest.** Missing, invalid or future timestamps do not count as healthy protection, and completed backups remain visible while another backup is active.
|
||||
- **Inventory failures stay local.** Unreadable backup responses do not publish misleading inventory, and a failure at one source does not invalidate independent results.
|
||||
- **PBS capacity alerts follow your policy.** Live polls apply datastore settings and preserve existing incidents when node status or measurements are incomplete.
|
||||
- **PBS capacity failures are visible.** Failed datastore reads no longer appear as healthy capacity in the backup dashboard.
|
||||
- **PBS History follows the actual machine.** Agent and Proxmox readings use corroborated host identity, rejecting ambiguous matches and withdrawing stale links.
|
||||
- **Proxmox navigation follows available data.** Tabs reflect observed capabilities, filters stay within their source, and node rows let you view that node's guests.
|
||||
- **Repeated site names stay separate.** Proxmox clusters with matching labels retain their own resources in live updates.
|
||||
- **Backup troubleshooting is safer.** Passive VM disk diagnostics avoid guest-agent commands, and guidance explains monitoring pauses without presenting them as proof that a backup is safe.
|
||||
|
||||
## TrueNAS, vSphere and Docker
|
||||
|
||||
- **TrueNAS CORE reporting works more consistently.** CORE 12 compatibility and native reporting improve memory, ARC, CPU, disk and network readings, including temperature history.
|
||||
- **TrueNAS polling recovers from silent peers.** Requests have bounded waits, idle sessions are maintained, and ended subscriptions no longer leave reads stalled.
|
||||
- **TrueNAS missing data stays distinct from zero.** Empty app statistics and incomplete reporting no longer become misleading measurements or synthetic CPU temperatures.
|
||||
- **TrueNAS incidents retain useful severity.** Emergency and actionable NOTICE alerts reach incident delivery, informational messages stay out, and finished replication jobs are recognised.
|
||||
- **Connection tests stay separate from monitoring.** TrueNAS probe results do not imply successful polling, while saved TrueNAS and vSphere connections retain runtime connection alerts.
|
||||
- **vSphere JSON API compatibility improves.** Calls support vSphere 8.0.3 requirements, tolerate absent values and expose more useful API fault details.
|
||||
- **Container metrics are more accurate.** Host and Docker CPU sampling remain independent, Podman CPU uses advancing interval readings, and Docker storage retains observed values.
|
||||
- **Docker image checks avoid false updates.** All local image digests count as current, and invalid registry responses are rejected.
|
||||
- **Container updates show confirmed outcomes.** Actions remain pending until results are known, require consistent running and health checks, and retain their age across restarts.
|
||||
|
||||
## Install, updates and agents
|
||||
|
||||
- **Automatic updates require a choice.** New installations ask before enabling unattended updates, and manual version changes preserve your existing preference.
|
||||
- **Update checks reflect installability.** Pulse checks for the correct server archive and ARM build variant, with more reliable handling of large release listings and unknown dates.
|
||||
- **Update progress stays responsive.** Quiet update streams trigger status checks, and Pulse confirms server readiness before reloading the dashboard.
|
||||
- **Successful updates report success.** Updater cleanup no longer causes a completed automatic update to exit with an error.
|
||||
- **FreeBSD installation is more dependable.** Checksum verification works without coreutils, and agent logs are collected on FreeBSD and pfSense.
|
||||
- **Agent continuity improves.** Credentials survive monitoring reloads, live state takes priority over old enrolment records, and removed agents remain blocked from re-enrolling.
|
||||
- **Windows host detection recovers.** Agents handle MachineGuid values with braces when collecting host information.
|
||||
- **Peer sensor collection is optional.** Proxmox agents can skip SSH sensor collection from cluster peers.
|
||||
- **Disconnected commands fail promptly.** Typed action requests stop waiting when their agent session drops, and stale action responses no longer overwrite current review state.
|
||||
- **Setup instructions are clearer.** First-source guidance explains what agents add, offers a container-console route to the setup token and separates monitoring access from command permissions.
|
||||
|
||||
## Pulse Pro, AI and hosted
|
||||
|
||||
- **Patrol shows its weekly work.** The Activity page summarises checks, findings, actions and estimated spend, with clear notices when retained history covers less than a week.
|
||||
- **Weekly summaries can arrive by email.** Settings > Reporting lets you schedule a Patrol summary with a weekday, time, timezone and recipients.
|
||||
- **Model costs are easier to compare.** Patrol setup previews estimated spending across schedules and clearly explains when an exhausted budget pauses analysis.
|
||||
- **Cached AI usage is counted.** Anthropic cache reads and writes contribute to usage, budgets and projections, including fully cached runs.
|
||||
- **Patrol findings are less repetitive.** Findings that mirror alerts are folded together, flapping is grouped, and lasting decisions remain visible.
|
||||
- **AI explanations retain context.** Patrol and alert actions can start scoped explanations, with resource history, action outcomes and uncertainty carried through investigations.
|
||||
- **Assistant access is more consistent.** Monitored targets remain available without command access, saved Patrol objectives recover, and unnecessary continuation prompts are removed.
|
||||
- **Provider configuration works better.** Ollama Basic Auth settings are restored, GPT-5 requests use supported completion limits, and stable prompt prefixes can be cached.
|
||||
- **AI controls reflect availability.** Patrol navigation is hidden when AI is off, and setup blocks leave existing attention items visible.
|
||||
- **Hosted clients survive upgrades more reliably.** Provider-managed client recovery, licence refresh and tenant cleanup respect the owning provider.
|
||||
- **Hosted billing reflects applied plans.** Provider-hosted platforms can buy or renew their own licence, and billing returns confirm the running plan before announcing success.
|
||||
- **Organisation settings are clearer.** Owners see the settings their organisation supports, hosted workspaces can create agent install tokens, and Relay labels and upsells are removed.
|
||||
|
||||
## Monitoring and service health
|
||||
|
||||
- **Larger installations do less repeated work.** Agent refreshes, resource lookups and dashboard broadcasts reduce copying and repeated processing.
|
||||
- **Metric storage is more resilient.** History handles startup write locks, retries, rollups and shutdown more reliably, with fewer duplicate samples across storage tiers.
|
||||
- **Samples retain their measurement time.** Unified metrics use source observation times, and used-memory history expires within its configured retention window.
|
||||
- **Statistics stay responsive.** Statistics reads are isolated from heavier history queries, and metric synchronisation batches database writes.
|
||||
- **Service checks are bounded.** Local health probes support IPv4 and IPv6 without waiting indefinitely, and cancelled reloads release waiting callers.
|
||||
- **Resource linking avoids false matches.** Host-local addresses are excluded from automatic matching, while explicit host links and custom bridge information are preserved.
|
||||
- **Linked memory remains available.** Proxmox VMs recover correlated agent memory, and LXC monitoring prefers linked agent readings.
|
||||
- **Saved platforms keep polling.** Missing organisation metadata no longer prevents monitoring configured connections.
|
||||
|
||||
## Security
|
||||
|
||||
- **Setup commands expose fewer secrets.** Proxmox bootstrap and agent tokens stay out of copied commands and hosted setup downloads, with private credential entry.
|
||||
- **Notification diagnostics mask credentials.** Webhook, Slack, Discord, Telegram, ntfy and Apprise diagnostics redact more credential forms and private payloads.
|
||||
- **Diagnostics exports redact more detail.** Downloads mask nested guest identities, token references, mount paths and upstream response details, with guidance to review files before sharing.
|
||||
- **History respects access changes.** Denied readings are withdrawn, and cached responses or drawer fallbacks cannot substitute data from another organisation.
|
||||
- **TLS choices remain yours.** Explicit certificate-verification settings survive Proxmox registration and consolidation.
|
||||
- **Login recovery preserves configuration.** Local authentication recovery and rotation retain authentication state and system preferences, with admin permissions synchronised during setup.
|
||||
- **SSO mappings are more compatible.** OIDC handles mixed key sets, and settings preserve group mappings containing spaces.
|
||||
- **Security dependencies are updated.** Go cryptography and frontend dependencies include fixes for newly identified vulnerabilities.
|
||||
|
||||
## Other improvements
|
||||
|
||||
- **Telemetry is a setup choice.** Consent moves into onboarding, and test or development builds avoid production telemetry reporting.
|
||||
- **Recovery help is more practical.** Updated guides explain data-preserving removal, configuration backup limits, trusted proxy headers and safe incident reporting.
|
||||
- **The demo is easier to enter.** Visitors are signed in automatically, and the banner leads directly to installation instructions.
|
||||
|
||||
## Before you upgrade
|
||||
|
||||
- Windows Unified Agent binaries are not Authenticode-signed while SignPath remains unavailable and may show an Unknown Publisher warning. Verify downloads with the published checksums and detached signatures.
|
||||
- Pulse Mobile remains compatible. This candidate does not require a companion mobile release.
|
||||
- The rollback target is stable `v6.4.5`. On systemd and Proxmox LXC installs, use this command only when `/bin/update` was installed by the Pulse server installer: `sudo /bin/update --version v6.4.5` to return to the previous stable release. For Docker Compose, pin `rcourtman/pulse:6.4.5` and recreate the container.
|
||||
@@ -0,0 +1,5 @@
|
||||
{
|
||||
"schema_version": 1,
|
||||
"captures": [],
|
||||
"decision": "Zero captures: the notes describe the changes in text."
|
||||
}
|
||||
@@ -0,0 +1,324 @@
|
||||
# Pulse v6.5.0-rc.1
|
||||
|
||||
This changelog describes the changes since `v6.4.5`. Stable `v6.4.5` remains the rollback target.
|
||||
|
||||
## Changes since v6.4.5
|
||||
|
||||
- Restore backup-alert protection and keep guest, backup and Diagnostics evidence truthful (#2463)
|
||||
- Cut 6.5 and later releases from main (#2468)
|
||||
- Preserve guest memory across repeated backup locks (#2462)
|
||||
- Keep Proxmox guest checks single-attempt and Diagnostics downloads safer (#2461)
|
||||
- Keep backup dates honest and saved platform monitoring resilient (#2460)
|
||||
- Keep alerts and backup protection honest when telemetry is unusable (#2459)
|
||||
- Keep guest details and History trustworthy, and clarify alerting limits (#2457)
|
||||
- Wait for the real token prompt in the PTY install test (#2458)
|
||||
- Keep guest readings truthful, expire memory history and improve safe setup guidance (#2456)
|
||||
- Keep guest disk freshness visible and count cached Patrol usage (#2455)
|
||||
- Report advancing Podman CPU and keep TrueNAS probe results separate from polling (#2454)
|
||||
- Open guest details on the first tap and clarify container-update recovery (#2453)
|
||||
- Make monitoring, notification results and recovery guidance safer and more truthful (#2429)
|
||||
- Measure route-normaliser allocations in an isolated child (#2451)
|
||||
- Record a browser pass for the merged stacked disk bar model (#2450)
|
||||
- Warn on inherited npm advisories when the lockfile is unchanged (#2442)
|
||||
- Show the last backup in Proxmox coverage and keep its tables whole on phones (#2448)
|
||||
- Account Anthropic prompt-cache usage instead of dropping it (#2431)
|
||||
- Proxmox and Docker page passes: storage words, first paint, table fits (#2449)
|
||||
- Carry Kubernetes job times into the controller row expansion (#2434)
|
||||
- Fit metric bar labels by measured glyph widths, not character count (#2438)
|
||||
- Fit storage table values on phones and move reasons to the drawer (#2433)
|
||||
- Keep the Patrol decision header pinned while a long decision scrolls (#2445)
|
||||
- Restore intended opacity classes Tailwind 3 never rendered (#2444)
|
||||
- Delete the retired shared browser receipt (#2437)
|
||||
- Record browser receipts per commit so frontend PRs stop conflicting (#2436)
|
||||
- Move frontend to Tailwind CSS 4 and jscpd 5 to drop braces (#2443)
|
||||
- Restore bounded alert delivery and evidence-first TrueNAS troubleshooting (#2428)
|
||||
- Let table callers own the cell padding they ask for (#2427)
|
||||
- Keep storage History and TrueNAS polling responsive and correctly attributed (#2420)
|
||||
- Offer the Patrol weekly summary as a scheduled report type (#2423)
|
||||
- Bound audit overhead and preserve the published release source mapping (#2412)
|
||||
- Show Kubernetes job times as an age, not a raw timestamp (#2411)
|
||||
- Keep mock replication jobs on schedule and off their own node (#2410)
|
||||
- Stop Proxmox tables clipping the values users came for (#2409)
|
||||
- Show and inspect single stored History readings (#2405)
|
||||
- Add a non-publishing Apple signing and notary preflight (#2403)
|
||||
- Keep disk temperatures visible without extended SMART attributes (#2401)
|
||||
- Show failed History refreshes without hiding the last successful readings (#2397)
|
||||
- Keep History readings on target and accessible by keyboard (#2393)
|
||||
- Show storage capacity truthfully and keep PBS diagnostics safe (#2392)
|
||||
- Keep TrueNAS polling responsive and disk health rows current (#2387)
|
||||
- Remove the echo-only release joins and trim unused artifact retention (#2374)
|
||||
- Keep Apprise diagnostics private and signed setup instructions fail closed (#2385)
|
||||
- Pin setup-node caching off in the dependency advisory watch (#2383)
|
||||
- Restore TrueNAS CORE metrics and keep queued alerts within quiet hours (#2379)
|
||||
- Keep quiet hours correct across clock changes and clarify TrueNAS diagnostics (#2377)
|
||||
- Keep healthy RAID spares from triggering critical alerts and preserve audit recovery evidence (#2373)
|
||||
- Keep update alerts steady and make setup and troubleshooting clearer (#2368)
|
||||
- Watch main and every active release line for new frontend advisories daily (#2366)
|
||||
- Keep PBS setup safety checks aligned with secure installation (#2367)
|
||||
- Make setup and administration safer and show sparse History honestly (#2362)
|
||||
- Rehearse install, upgrade and rollback between published releases (#2363)
|
||||
- Keep missing TrueNAS readings out of live usage and History (#2360)
|
||||
- Make TrueNAS troubleshooting safer and History timelines accurate (#2359)
|
||||
- Keep alert acknowledgements through long-running incidents (#2358)
|
||||
- Keep History readings correctly attributed and continuing alerts visible (#2357)
|
||||
- Reduce alert checkpoint churn and clarify ZFS monitoring (#2356)
|
||||
- Keep History charts usable when refreshes fail (#2355)
|
||||
- Keep the update progress modal moving when the update stream goes quiet (#2347)
|
||||
- Keep host History correctly scoped and make startup health checks reliable (#2341)
|
||||
- Raise brace-expansion and DOMPurify past current advisories (#2349)
|
||||
- Keep Docker update outcomes truthful and PBS History tied to the right host (#2339)
|
||||
- Record Backup Server CPU and memory history after successful polls (#2338)
|
||||
- Keep action reviews safe and notify when incidents continue (#2335)
|
||||
- Prevent duplicate alert delivery and make pending actions safer to review (#2334)
|
||||
- Make update recovery safer and restore alert and resource history (#2323)
|
||||
- Validate merged integration ranges against in-range browser receipts (#2330)
|
||||
- Let the RC attester read the lab's root-owned evidence (#2329)
|
||||
- Give RC qualification a local host-interface canary (#2328)
|
||||
- Calibrate internal/api shard weights to CI and record per-test seconds (#2326)
|
||||
- Retire Relay labels and upsells inside Pulse (#2325)
|
||||
- Balance internal/api race shards by measured run time (#2324)
|
||||
- Shard internal/api race tests and drop probation E2E from PRs (#2322)
|
||||
- Verify AI-off navigation and clearer setup guidance (#2316)
|
||||
- Restore same-RC release verification and fix Synology temperature and PBS backup age (#2315)
|
||||
- Match setup handoff E2E to the corrected first-source copy (#2314)
|
||||
- Stop describing Patrol modes on Actions while AI is off (#2313)
|
||||
- Hide Patrol navigation when AI is off and rename Updates (#2312)
|
||||
- Clarify when first-source setup can show host sensors (#2309)
|
||||
- Soak release candidates for 24 hours before stable (#2307)
|
||||
- Use plain words on the first-source setup screen (#2303)
|
||||
- Make update readiness and history tests reliable (#2304)
|
||||
- Show container-console route for the setup token (#2302)
|
||||
- Fail typed dispatches fast when their agent session drops (#2300)
|
||||
- Give update check test fixtures the exact runtime archive (#2297)
|
||||
- Keep the non-procfs agent recovery test off the host's /proc (#2298)
|
||||
- Make update offers, backup badges, and Docker alerts more reliable (#2293)
|
||||
- Keep queued alerts safe and restore reliable update checks (#2288)
|
||||
- Report update channel and last update check outcome in telemetry (#2285) (#2286)
|
||||
- Stream release metadata so update checks see stable releases (#2282) (#2284)
|
||||
- Keep Vitest and coverage updates compatible (#2279)
|
||||
- Restore dependable disk alerts and linked guest storage (#2270)
|
||||
- Name Proxmox in the README headline for search (#2265)
|
||||
- Record the revised registry benchmark result (#2261)
|
||||
- Bound large-estate refresh and storage navigation work (#2260)
|
||||
- Preserve alert lifecycle truth and safe pinned updates (#2259)
|
||||
- Stop timing the per-segment route normaliser (#2258)
|
||||
- Confirm paid provider plans from the refreshed licence (#2256)
|
||||
- Stop timing the one-cycle NormalizeRoute root fast path (#2255)
|
||||
- Keep the provider portal up when the MSP licence lapses (#2253)
|
||||
- Refresh the resource store at most once per 2s of agent reports (#2242)
|
||||
- Retire ChangeDetector's dead change-detection path (#2252)
|
||||
- Let provider MSP upgrade.sh move an install to a new release (#2251)
|
||||
- Show the real Pulse mark in the branding header preview (#2249)
|
||||
- Reuse the host continuity overlay and copy only hosts for live lookups (#2245)
|
||||
- Space websocket state broadcasts at least 2s apart per audience (#2246)
|
||||
- Audit agent config fetches only when they carry new information (#2244)
|
||||
- Keep the provider MSP install proof out of client slots (#2239)
|
||||
- Remove ChangeDetector's unjoined legacy history writer (#2234)
|
||||
- Compare the renewed licence expiry in UTC in its test (#2232)
|
||||
- Make agent report refreshes linear and clone the registry once (#2235)
|
||||
- Verify provider network isolation before Pulse releases (#2231)
|
||||
- Make the time-major metrics index the unique identity index (#2230)
|
||||
- Join queued incident saves before test cleanup (#2229)
|
||||
- Stop copying the resource registry on every read-state lookup (#2228)
|
||||
- Reduce CPU work when resource snapshots are accepted (#2225)
|
||||
- Keep provider MSP client routes online after upgrades and isolated across providers (#2224)
|
||||
- Give provider MSP operators accurate plan and client-limit guidance (#2223)
|
||||
- Point provider MSP setup at the portal Plan tab to buy (#2221)
|
||||
- Confirm purchased MSP plans only after the new licence is active (#2222)
|
||||
- Provider MSP setup starts a working control plane after client recovery (#2220)
|
||||
- Keep provider MSP clients running across upgrades (#2216)
|
||||
- Honor shortened provider MSP paid-licence periods (#2219)
|
||||
- Let a provider-hosted platform buy, renew and see its own plan (#2215)
|
||||
- Mint agent install tokens in hosted client workspaces (#2209)
|
||||
- Guard orphaned-backup opt-out and TrueNAS disk-write boundaries (#2211)
|
||||
- Judge client health from Docker when status runs out of band (#2210)
|
||||
- Offer org owners the settings their own org's routes serve (#2208)
|
||||
- Reduce resource-view read contention and verify TrueNAS metric targets (#2206)
|
||||
- Let stable promotion carry test-only and installer pin changes (#2204)
|
||||
- Keep release and provider-MSP frontend builders on the updated Node 24 image (#2202)
|
||||
- feat(agent): allow disabling Proxmox peer sensor collection (#2177)
|
||||
- Make releasing the default and define an achievable soak (#2200)
|
||||
- Keep PBS History on its host target when the connection uses an IP or alias (#2198)
|
||||
- Make Profile M row-windowing checks reliable under load (#2195)
|
||||
- Keep the Backups History chart on its host series (#1723) (#2194)
|
||||
- Bump golang in /.devcontainer (#2188)
|
||||
- Keep the eight governed release action pins out of Dependabot version updates (#2193)
|
||||
- Correct stable-patch rollback metadata and qualify the managed Go dependency group (#2185)
|
||||
- Serialize AI knowledge-store saves and refresh the portal bundler (#2183)
|
||||
- Refresh governed container base-image digests (#2182)
|
||||
- Capture agent logs on FreeBSD/pfSense and fix the installer log hint (#2180)
|
||||
- Keep the resource drawer's selected tab across data refreshes (#2178)
|
||||
- Deliver persisted alert notifications that are due at startup (#2175)
|
||||
- Fix TrueNAS CORE CPU and memory telemetry over legacy REST (#2172)
|
||||
- Deliver critical alert escalations after warning notifications (#2170)
|
||||
- Keep TrueNAS telemetry available when optional graphs fail (#2168)
|
||||
- Keep copied installs usable and prevent duplicate disk rows (#2166)
|
||||
- Group resolved alert notifications to prevent restart delivery bursts (#2164)
|
||||
- Make guest filesystem mount paths readable (#2162)
|
||||
- Expand PBS History refresh coverage across reported host layouts (#2159)
|
||||
- Keep PBS datastore History attached during refresh (#2156)
|
||||
- Fix PBS host history correlation and alert-card footer alignment (#2154)
|
||||
- Fix spurious PBS backup alerts and Proxmox LXC memory; qualify release-line RCs (#2152)
|
||||
- Bump alpine from 3.24 to 3.24 in /tests/integration (#2151)
|
||||
- Recover saved Patrol objectives without observers (#2147)
|
||||
- Fix node Verify SSL setting reverting after re-registration (#2140) (#2144)
|
||||
- Ship reviewed frontend dependency security updates (#2141)
|
||||
- Ship reviewed maintenance fixes for metrics, monitoring, TrueNAS, VMware and the updater (#2134)
|
||||
- Clarify contractor administration of customer-owned licences (#2120)
|
||||
- Fix CSRF test initialization for release preflight (#2116)
|
||||
- chore(portal): patch Vitest security dependency (#2114)
|
||||
- build(deps-dev): bump typescript from 5.9.3 to 7.0.2 in /tests/integration (#2103)
|
||||
- build(deps-dev): bump @types/node from 20.19.24 to 26.5.1 in /tests/integration (#2102)
|
||||
- build(deps): bump golang in /.devcontainer (#2105)
|
||||
- Keep agent credentials consistent across monitor reloads (#2108)
|
||||
- Preserve settings during security setup and bind stable promotion to observed source (#2091)
|
||||
- Cancel superseded pull request validation (#2089)
|
||||
- Preserve warning-level alert notification preferences (#2087)
|
||||
- fix: preserve disk identity and tolerate TrueNAS alert argument shapes (#2082)
|
||||
- Keep qualified releases safe through skipped checks and cancellation (#2074)
|
||||
- Recognise canonical vSphere 8.0.3 without breaking connection checks (#2072)
|
||||
- Show a node's guests from the Proxmox overview row (#2065)
|
||||
- Isolate release preparation and qualification on hosted VMs (#2063)
|
||||
- Retain the reviewed release-version reuse guard (#2060)
|
||||
- Prevent published release versions from being reused (#2058)
|
||||
- fix(release): qualify candidates before public version writes (#2056)
|
||||
- fix(metrics): isolate statistics reads from history load (#2054)
|
||||
- Show every disk mount without a hidden inner scrollbar (#2052)
|
||||
- Stop registry error pages from appearing as Docker image updates (#2049)
|
||||
- Reduce unnecessary alert-state disk writes (#2046)
|
||||
- Route Pulse 6.4.5 reliability releases through the maintained release line (#2045)
|
||||
- Keep diagnostic source builds out of release update flows (#2034)
|
||||
- Keep monitoring snapshots consistent during state resets (#2041)
|
||||
- Restore authenticated Ollama settings and make performance checks reproducible (#2039)
|
||||
- Make lifecycle controls available when planning is ready (#2037)
|
||||
- diagnostic: run hosted lifecycle fixture on DeepSeek V4.1 Flash (#2036)
|
||||
- Verify notification recovery across retries and restarts (#2033)
|
||||
- Make installed notification recovery testable across retries and restarts (#2032)
|
||||
- Preserve live agent state over saved enrollment history (#2031)
|
||||
- Fix manual update freshness and unknown release dates (#2030)
|
||||
- Safely verify hosted Assistant lifecycle planning (#2029)
|
||||
- Prevent stale alert escalations after alert state changes (#2028)
|
||||
- Align release policy with immutable publication (#2027)
|
||||
- Prioritize dependable release delivery (#2026)
|
||||
- Prevent cleared alerts returning and restore admin settings access (#2025)
|
||||
- Keep automatically acknowledged alerts acknowledged after brief recovery (#2024)
|
||||
- Preserve the reviewed agent connection ordering history (#2023)
|
||||
- Keep agent commands ordered during connection and verify Gemini reboot planning (#2021)
|
||||
- Preserve import previews and PBS history while updating dependencies (#2019)
|
||||
- Keep Prometheus metrics exports compatible after dependency updates (#2017)
|
||||
- Keep Docker inventory compatible with Moby API 1.56 (#2015)
|
||||
- Make MSP evaluation guidance accurate for current releases (#2014)
|
||||
- Honor PBS capacity thresholds and keep dependency upgrades safe (#2012)
|
||||
- Add regression coverage for standalone Proxmox connection renames (#2011)
|
||||
- Make power sensor checks deterministic (#2008)
|
||||
- Prevent unrelated Proxmox providers from sharing agent identity (#2007)
|
||||
- Keep organization controls visible and deliver critical alert escalations (#2005)
|
||||
- Keep explicitly included disks visible and strengthen managed browser checks (#2004)
|
||||
- Protect disk monitoring and organisation access boundaries (#2003)
|
||||
- Make Settings startup failures actionable without weakening checks (#2002)
|
||||
- ci: avoid duplicate E2E artifact uploads (#2001)
|
||||
- Protect webhook credentials in mixed-case diagnostics (#1999)
|
||||
- Capture exact timing for release history performance checks (#1998)
|
||||
- Explain safe Proxmox access for pending update checks (#1997)
|
||||
- Improve update guidance, webhook parity, and frontend security (#1995)
|
||||
- Make API stress checks easier to diagnose and isolate (#1994)
|
||||
- Retain timing evidence when release stress tests fail (#1993)
|
||||
- Keep release reconciliation moving through transient GitHub errors (#1992)
|
||||
- Keep tenant resource requests responsive during cold store setup (#1991)
|
||||
- Keep TrueNAS sessions alive and refresh on a steady cadence (#1990)
|
||||
- Keep transient webhooks retrying and explain destination server failures (#1988)
|
||||
- docs: explain release channels and qualification (#1987)
|
||||
- Keep notification recovery responsive and issue follow-up accurate (#1986)
|
||||
- Honor service recovery timing and protect saved alert settings (#1985)
|
||||
- Strengthen alert retry and incident refresh regression coverage (#1984)
|
||||
- Document standing maintainer reply and release authority (#1983)
|
||||
- Make release rehearsals dependable and strengthen alert regression coverage (#1982)
|
||||
- Strengthen alert recovery, backup-setting, and browser lifecycle evidence (#1981)
|
||||
- Clarify notification retry limits and recovery validation (#1980)
|
||||
- Verify reliable alert recovery across restarts (#1979)
|
||||
- Keep alert retries and history recovery dependable (#1976)
|
||||
- Keep alert history requests isolated and preserve canonical history (#1974)
|
||||
- Record incident history qualification and delivery (#1975)
|
||||
- Preserve canonical incident history and Assistant handoffs (#1973)
|
||||
- Prevent stale monitoring from looking healthy (#1972)
|
||||
- Strengthen recurring incident and guest-memory regression coverage (#1969)
|
||||
- Keep recurring incidents distinct and clear withdrawn Proxmox memory (#1967)
|
||||
- Protect alert credentials and preserve accurate recovery state (#1959)
|
||||
- Publish reviewed release snapshots independently of branch tips (#1965)
|
||||
- Record verified Patrol and Assistant qualification delivery (#1963)
|
||||
- Preserve canonical Patrol planning and outcome continuity (#1960)
|
||||
- Protect alert settings navigation and threshold edits from regressions (#1958)
|
||||
- Remove inferred Assistant continuation gates (#1957)
|
||||
- Keep Retry and Dismiss outcomes visible after toasts expire (#1953)
|
||||
- Use typed Proxmox runners and preserve cross-clock action evidence (#1955)
|
||||
- Protect notification explanations through holds and outages (#1949)
|
||||
- Preserve native filesystem evidence and Patrol action history (#1951)
|
||||
- Fix release smoke workspace identity on rootless Docker (#1948)
|
||||
- Stop permanent alert retries and keep delivery and backup evidence truthful (#1946)
|
||||
- Preserve diagnostic evidence and Patrol outcome truth (#1935)
|
||||
- Link the demo banner back to the install steps (#1945)
|
||||
- Keep delivery warnings current and alert checks reliable (#1944)
|
||||
- Keep alert delivery status from reverting after newer checks (#1942)
|
||||
- Sign demo-mode visitors in instead of showing them demo/demo (#1941)
|
||||
- Prepare a forward 6.4.4 beta and preserve storage crash diagnostics (#1939)
|
||||
- Show alert dispatch honestly and guard responsive startup (#1938)
|
||||
- Keep host links stable and storage alerts dependable through restart (#1932)
|
||||
- fix(assistant): preserve tool evidence identity (#1934)
|
||||
- Keep storage alerts dependable and TrueNAS setup checks accurate (#1931)
|
||||
- perf(api): avoid rune decoding in route label checks (#1929)
|
||||
- fix(patrol): preserve evidence across history and access failures (#1928)
|
||||
- Protect alert recovery receipts across gaps and crashes (#1927)
|
||||
- fix(ai): preserve canonical evidence through diagnostic workflows (#1920)
|
||||
- Protect alert continuity and TrueNAS investigation coverage (#1926)
|
||||
- Keep persisted alerts from breaking threshold checks (#1924)
|
||||
- Keep alert recovery visible and storage thresholds stable after outages (#1923)
|
||||
- Keep alert delivery and release checks trustworthy (#1918)
|
||||
- fix(patrol): preserve refusals and expose measured summary evidence (#1917)
|
||||
- Repair resource evidence and Patrol readiness gates (#1916)
|
||||
- Keep alerts accurate through outages, restarts, and genuine recovery (#1909)
|
||||
- Preserve investigation evidence and Proxmox temperatures (#1914)
|
||||
- Preserve Patrol finding volume beyond the history cap (#1912)
|
||||
- Start scoped explanations from Patrol and alert actions (#1911)
|
||||
- Keep alerts, navigation, and release checks trustworthy (#1904)
|
||||
- Leave release cadence to maintainer judgment (#1907)
|
||||
- Keep alerts, monitoring history, and navigation reliable through interruptions (#1900)
|
||||
- Keep monitoring, alerts, and release qualification dependable (#1896)
|
||||
- Stop retrying notification failures that cannot succeed (#1879)
|
||||
- Order the Patrol HTTP observer test against its health-lease write (#1886)
|
||||
- Bound the npm audit retry budget by wall clock (#1889)
|
||||
- Classify every 0.0.0 sentinel build as a development build (#1884)
|
||||
- Give the Frontend job room for the npm audit retries (#1888)
|
||||
- Separate a real advisory from an unreachable npm audit endpoint (#1885)
|
||||
- Record the Patrol investigation rate as an invalid metric (#1883)
|
||||
- Stop test binaries reporting to the production telemetry endpoint (#1878)
|
||||
- Classify notification failures from senders, not error prose (#1877)
|
||||
- Make telemetry consent a setup choice and retire the payload banner (#1874)
|
||||
- Make Helm Pages convergence idempotent on immutable chart releases (#1871)
|
||||
- Qualify release candidates against the branch that owns their version (#1870)
|
||||
- Improve accessibility, TrueNAS readiness, and release reliability (#1862)
|
||||
- Capture the metrics startup hook per store to end a test flake (#1868)
|
||||
- Send max_completion_tokens to GPT-5 models and learn it from the API (#1867)
|
||||
- Add schema v17 Patrol provider, cost, autonomy, and outcome telemetry (#1864)
|
||||
- Email the Patrol weekly summary as a report schedule kind (#1865)
|
||||
- Restore Core E2E on main: link navigation, Patrol controls, badge contrast (#1863)
|
||||
- Stop background git from racing test_docs_mirror temp cleanup (#1859)
|
||||
- Show what Patrol did this week on the Patrol page (#1860)
|
||||
- Fix the release preflight worker and Helm publish, and map v6.4.3 to release/v6.4 (#1840)
|
||||
- Fix empty-slot alerts and make keyboard navigation accessible (#1852)
|
||||
- Add the Patrol weekly digest endpoint (#1856)
|
||||
- Guide the Patrol model choice with cost preview and budget pause (#1853)
|
||||
- Fold alert-mirroring findings, collapse flapping, surface lasting decisions (#1848)
|
||||
- Make release checks reliable and dialogs accessible (#1842)
|
||||
- Fix rootful qualification readiness (#1839)
|
||||
- Fix isolated rootful attester loading (#1834)
|
||||
- Describe how changes land and ship, and hold human commits to the standard (#1828)
|
||||
|
||||
## Release Metadata
|
||||
|
||||
- Version: `v6.5.0-rc.1`
|
||||
- Previous stable: `v6.4.5`
|
||||
- Rollback target: `v6.4.5`
|
||||
- Rollback command: `sudo /bin/update --version v6.4.5`
|
||||
- Promotion path: exact-SHA single-build release candidate from `main`
|
||||
- Windows signing decision: prereleases publish checksum- and detached-signature-verified Windows agents without Authenticode while SignPath remains unavailable. Windows Unified Agent binaries may display an Unknown Publisher warning
|
||||
- Mobile decision: `no-mobile-impact`. No governed mobile route, payload, Relay, pairing, approval, push or onboarding contract changed from `v6.4.5`, so no companion mobile build or store rollout is required
|
||||
@@ -4,8 +4,8 @@ This guide covers practical upgrade steps for existing Pulse installs moving to
|
||||
|
||||
For the current v6 release candidate packet, see:
|
||||
|
||||
- `docs/releases/RELEASE_NOTES_v6.4.3-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.4.3-rc.1.md`
|
||||
- `docs/releases/RELEASE_NOTES_v6.5.0-rc.1.md`
|
||||
- `docs/releases/V6_CHANGELOG_v6.5.0-rc.1.md`
|
||||
|
||||
For the current stable v6 packet, see:
|
||||
|
||||
|
||||
@@ -6,7 +6,7 @@ set -euo pipefail
|
||||
|
||||
SCRIPT_DIR="$(CDPATH= cd -- "$(dirname -- "$0")" && pwd)"
|
||||
DOCKER_IMAGE_REPO="${DOCKER_IMAGE_REPO:-rcourtman/pulse}"
|
||||
CANONICAL_DEFAULT_PULSE_VERSION="6.4.3-rc.1"
|
||||
CANONICAL_DEFAULT_PULSE_VERSION="6.5.0-rc.1"
|
||||
|
||||
resolve_default_pulse_version() {
|
||||
if [ -n "${PULSE_IMAGE_VERSION:-}" ]; then
|
||||
|
||||
Reference in New Issue
Block a user