Repository navigation
🐛 fix: clear resolved upgrade delay warnings - #15747
Conversation
Signed-off-by: Andrew Anderson <andy@clubanderson.com>
|
/lgtm |
|
@clubanderson: you cannot LGTM your own PR. DetailsIn response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes-sigs/prow repository. |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: clubanderson The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
Open PR overlap checkBase branch: Conflicts are checked with every PR merged onto the current tip of No conflicting open PRs or same-file overlaps were found. |
There was a problem hiding this comment.
Code looks correct to me; one operator-doc gap below.
What I verified (read at head aa824df):
src/pkg/hub/spoke/upgrade_history.go:8-30only relabels rows withOutcome == delayedorTrigger == window && Outcome == missed.UpgradeTriggerWindowis written solely bywindowMissed(src/pkg/hub/upgrade_history.go:190-209) and is documented as "not an attempt at all" (spoke/upgrade_progress.go:58-60), so no genuinely failed attempt gets relabelled as a hold. Persisted history is untouched (recent()returns the filtered copy,byIDstays raw).src/pkg/hub/upgrade_progress.go:141-142: the live-hold branch now also requiresin.HoldAt.After(last.EndedAt).AutoUpgradeHoldAtis always set alongsideAutoUpgradeHoldReason(upgrade_schedule.go:290-291) and cleared with it (saas_upgrade.go:929-930,saas_bulk.go:396-397), so the zero-HoldAtcase cannot suppress a real hold.heartbeat_upgrade_policy.go:66ships the filteredrecent()to the spoke, which runsUpgradeHistoryForDisplayagain indashboard/upgrade_progress.go:66-70andapi_version.go:98-104— the function is idempotent so the double pass is harmless.- Both pill tables (
static-src/js/version.js,hub/assets/dashboard.html) gaineddelayedwith the amber draining styling; the hubhistorybadge now counts only non-window misses newer than the lastdone.
Docs-currency (low): src/docs/operator-reference.md:1348 still lists missed as covering "the schedule gate's hold reason while the window is still open", :1352 says outcome (done / missed), and :1354 says the hub pill appears for … verifying and missed and the spoke chip shows Missed window while a hold is active. All three now describe the new delayed state/outcome. Worth a short edit in this PR — note #15731 (split the operator reference) is open and would move these lines, so whichever lands second needs a rebase there.
— hive: agent=reviewer-2 backend=copilot model=claude-fable-5.1 copilot=1.0.88
There was a problem hiding this comment.
Read the diff against the tree at aa824df, plus the callers; this looks correct to me.
What I checked:
- Relabelling
window+missedrows todelayedis safe.UpgradeHistoryForDisplay(src/pkg/hub/spoke/upgrade_history.go:15-16) only rewrites rows whoseTrigger == UpgradeTriggerWindow, and that trigger is documented as "not an attempt at all" (spoke/upgrade_progress.go:58-60) and is written only bywindowMissed(hub/upgrade_history.go:203);start()never uses it. So real attempts that failed inside a window keepmissed. - Hub live-hold vs. history row agree.
windowMissedstamps the hold rowEndedAt = at(hub/upgrade_history.go:194,204), so once recorded the new guardin.HoldAt.After(last.EndedAt)(hub/upgrade_progress.go:142) is false and the pill falls through to the history branch, which emits the samedelayedstate,Reason,NextWindowAtand"delayed: …"detail (hub/upgrade_progress.go:154-168). No visible flicker between the two paths. - Audit log untouched. Only
recent()(hub/upgrade_history.go:75) applies the display filter;byID/persistLockedkeep the raw rows, andrecent()'s two callers (heartbeat_upgrade_policy.go:66,upgrade_progress.go:219) are both display/heartbeat paths. The spoke re-applies the same filter idempotently with its ownOutcome.CompletedAt, which is what clears a hold before the next heartbeat. dashboard.html'smissedcounter now excludestrigger === 'window'rows and anything older than the lastdone, matching the Go side.go build ./pkg/hub/... ./pkg/dashboard/is clean at the head;gofmtclean on the touched Go files.
Body claims match the diff (Delayed state, cleared holds, audit log preserved).
— hive: agent=reviewer backend=copilot model=claude-fable-5.1 copilot=1.0.88
|
merged by Hive trusted-author auto-merge: author @clubanderson holds role owner and GitHub permission admin on repo; CI green at 2448337. |
|
Thank you for your contribution! Your PR has been merged. We'd love to hear how your experience was: share feedback |
Summary
Coverage
Added shared history normalization, hub/spoke state, heartbeat policy, version API, and shipped JavaScript rendering regressions. Regenerated the spoke dashboard from source fragments. Pre-commit code review found no significant issues. Test/build/lint verdicts are delegated to CI per AGENTS.md.