Skip to content

fix(observe): keep platform health during a full online drain - #402

Merged
rldyourmnd merged 1 commit into
mainfrom
fix/online-drain-platform-health
Sep 4, 2026
Merged

fix(observe): keep platform health during a full online drain#402
rldyourmnd merged 1 commit into
mainfrom
fix/online-drain-platform-health

Conversation

@rldyourmnd

Copy link
Copy Markdown
Contributor

Summary

An authorized online drain of every compute member pages fleet_platform_unhealthy because uncovered-beyond-grace still zeros platform health. The listing is complete (empty is the truth), so gha_fleet_visibility_held_out_members correctly stays 0 and queue/assigned pages keep firing.

When every cluster member has a drain reason, uncovered-beyond-grace is expected capacity loss. Platform health stays up. A partial drain still fails health: an undrained sibling can still place work.

Verify

  • gofmt -l internal/fleetobserve
  • go vet ./internal/fleetobserve/
  • go test ./internal/fleetobserve/ ./cmd/gha-fleet-observer/

Held-out visibility still applies only to offline drain-marked members,
so queue and assigned pages keep firing. When every cluster member has
a drain reason, uncovered-beyond-grace is expected capacity loss and
must not zero gha_fleet_platform_healthy.
@rldyourmnd
rldyourmnd merged commit 7a218aa into main Sep 4, 2026
10 checks passed
@rldyourmnd
rldyourmnd deleted the fix/online-drain-platform-health branch September 4, 2026 22:39
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant