You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
Guard host recovery against changing Node resource versions #6364
Guarded storage-host recovery can leave application scheduling paused when an unrelated Node status update occurs during its safety checks. The release PATCH currently uses the resource version captured before those checks. Recovery also reports the same generic command failure for several distinct stages, making a safe resume harder to diagnose.
Steps to Reproduce
Start the existing host-recovery fixture in its saved returning phase.
Advance only the Node resource version during the native or peer safety observations.
Require the release PATCH to use the current resource version while retaining its identity preconditions.
Expected Behavior
Recovery uses a freshly verified version of the same owned Node immediately before releasing scheduling and before removing its completion marker. An uncertain write is accepted only through exact identity-bound readback; no second reboot or broad write retry is allowed. Full storage and database recovery remains required for completion.
Actual Behavior
Both Node mutations can use an older resource version. A rejected PATCH holds the cordon and the generic error does not identify the failed recovery stage. This code window is confirmed; its connection to the recent production hold is not yet proven.
Environment
The reviewed Platform host-memory recovery workflow.
Acceptance Criteria
Demonstrate the resource-version race with RED/GREEN fixture evidence.
Preserve native host, disk, boot, ownership, current-main and peer-copy checks.
Confirm ambiguous journal or scheduling writes with exact readback and one write only.
Keep private command output out of diagnostics.
Verify a real resumed recovery and application readiness separately from local tests.
#5806 - feat(storage): prepare hosts and temporary raw storage for V2 [merged]
📝 Issue Planner
Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.
Create Plan
🧪 Issue enrichment is currently in open beta.
You can configure auto-planning by selecting labels in the issue_enrichment configuration.
To disable automatic issue enrichment, add the following to your .coderabbit.yaml:
issue_enrichment:
auto_enrich:
enabled: false
💬 Have feedback or questions? Drop into our discord!
Production deployments and dependent sign-in delivery are blocked by unfinished storage-host recovery. This is a scoped rung-0 WIP exception under the consumer's immediate live-breakage rule; the fresh board census is already over its implementation ceiling. Work is limited to the recovery defect and its safety regression coverage, with ordinary review and production gates retained.
Priority rationale: Urgent, because the recovery hold blocks production delivery across applications. Service class: Expedite. End-to-end effort: Medium, including adversarial fixtures, CI, independent review, protected deployment and native resumed-recovery/application readback. First implementation start: 2026-11-15. The existing reviewed recovery is still queued; the proposed correction grants no direct cluster mutation, queue priority change or repeat reboot.
Activity
coderabbitai commentedon Nov 15, 2026
🔗 Related PRs
#5806 - feat(storage): prepare hosts and temporary raw storage for V2 [merged]
📝 Issue Planner
Check the box below or use the
@coderabbitai plancommand to generate an implementation plan and prompts that you can use with your favorite coding assistant.🧪 Issue enrichment is currently in open beta.
You can configure auto-planning by selecting labels in the issue_enrichment configuration.
To disable automatic issue enrichment, add the following to your
.coderabbit.yaml:💬 Have feedback or questions? Drop into our discord!
devantler commentedon Nov 15, 2026
Production deployments and dependent sign-in delivery are blocked by unfinished storage-host recovery. This is a scoped rung-0 WIP exception under the consumer's immediate live-breakage rule; the fresh board census is already over its implementation ceiling. Work is limited to the recovery defect and its safety regression coverage, with ordinary review and production gates retained.
Priority rationale: Urgent, because the recovery hold blocks production delivery across applications. Service class: Expedite. End-to-end effort: Medium, including adversarial fixtures, CI, independent review, protected deployment and native resumed-recovery/application readback. First implementation start: 2026-11-15. The existing reviewed recovery is still queued; the proposed correction grants no direct cluster mutation, queue priority change or repeat reboot.
chore: agent-claim #6364 nonce=740bb2024d7c01bb99a7f54891a3850a
chore: agent-claim #6364 nonce=c8b4c94af4d87c9bc7f83f8193cbc22f
chore: agent-claim #6364 nonce=00f80e4f7168be37695ae04a443f87a7
test(storage): share recovery write-response cases