Refactor BIOSVersionSetReconciler
#572
Conversation
Harmonize and simplify the reconciler code.
afritzler
force-pushed
the
enh/biosversionset
branch
from
41e9b8b to
8df5d0e
Compare
| if err := r.Get(ctx, req.NamespacedName, versionSet); err != nil { | ||
| return ctrl.Result{}, client.IgnoreNotFound(err) | ||
| } | ||
| log.V(1).Info("Reconciling biosVersionSet") |
There was a problem hiding this comment.
Should we also harmonize having these placeholder log messages across controllers? That we know, we expect it when certain CRD is being processed?
There was a problem hiding this comment.
I would say so, like the actually reconciliation of the object happens in the r.reconcile method. That is where the log belongs.
|
@coderabbitai review |
✅ Actions performedReview triggered.
|
WalkthroughRefactoring of the BIOSVersionSet controller internal logic, including variable renaming from biosVersionSet to versionSet, switching from updateStatus to patchStatus operations, reworking ownership-based retrieval methods, replacing helper functions with new variants, and updating watch configuration to use ownership-based patterns instead of event-based predicates. Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes
Pre-merge checks and finishing touches❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✨ Finishing touches
🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 0
🧹 Nitpick comments (2)
internal/controller/biosversionset_controller.go (2)
166-173: Status computed from stale versions list.The
versionslist is fetched at line 145, butensureBIOSVersionsForServersanddeleteOrphanBIOSVersionsmodify the cluster state before status computation at line 167. The computed status won't reflect newly created or just-deleted BIOSVersions.While the
Ownspredicate will trigger another reconciliation to correct this, the patched status may briefly show incorrect counts (e.g.,AvailableBIOSVersionexcludes just-created versions or includes just-deleted ones).Consider refetching versions after the create/delete operations for immediate accuracy:
Proposed fix
if err := r.patchBIOSVersionFromTemplate(ctx, log, &versionSet.Spec.BIOSVersionTemplate, versions); err != nil { return ctrl.Result{}, fmt.Errorf("failed to patch BIOSVersion spec from template: %w", err) } + // Refetch versions to reflect creates/deletes before computing status + versions, err = r.getOwnedBIOSVersions(ctx, versionSet) + if err != nil { + return ctrl.Result{}, fmt.Errorf("failed to refresh owned BIOSVersions: %w", err) + } + log.V(1).Info("Updating the status of BIOSVersionSet") status := r.getOwnedBIOSVersionSetStatus(versions)
331-345: Comment clarification.The comment on line 334 says "if the label has been removed", but this else-branch also handles cases where labels simply don't match the selector (not necessarily removed). Consider updating for clarity:
- } else { // if the label has been removed + } else { // if the Server no longer matches the selectorThe logic itself is correct—reconciling the set allows orphan BIOSVersion cleanup when a Server no longer matches.
📜 Review details
Configuration used: Organization UI
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (1)
internal/controller/biosversionset_controller.go(3 hunks)
🧰 Additional context used
🧬 Code graph analysis (1)
internal/controller/biosversionset_controller.go (3)
api/v1alpha1/biosversionset_types.go (3)
BIOSVersionSet(49-55)BIOSVersionSetStatus(21-34)BIOSVersionSetList(60-64)api/v1alpha1/biosversion_types.go (6)
BIOSVersionList(141-145)BIOSVersion(130-136)BIOSVersionStateInProgress(19-19)BIOSVersionStateCompleted(21-21)BIOSVersionStateFailed(23-23)BIOSVersionStatePending(17-17)api/v1alpha1/server_types.go (1)
ServerList(439-443)
🔇 Additional comments (10)
internal/controller/biosversionset_controller.go (10)
44-51: LGTM!The Reconcile entry point follows standard controller-runtime conventions. The variable rename to
versionSetis consistent throughout the refactor.
53-58: LGTM!Clean delegation pattern based on deletion timestamp.
60-97: LGTM!The deletion logic correctly ensures all owned BIOSVersions reach terminal state (Completed or Failed) before removing the finalizer. The switch to
patchStatusandPatchEnsureNoFinalizeraligns with best practices for conflict-free updates.
99-123: LGTM!Clean pattern for annotation propagation with appropriate early return when no BIOSVersions exist.
183-219: LGTM!The function correctly handles:
- Avoiding duplicate BIOSVersion creation via the
withBiosVersionmap- Name length limits with
GenerateNamefallback- Proper owner reference setup for the
Ownspredicate to work
221-240: LGTM!Good defensive handling by skipping
InProgressBIOSVersions to avoid interrupting active operations.
242-265: LGTM!Properly skips
InProgressversions to avoid mid-operation template changes. TheOperationResultNonefilter for logging reduces noise from no-op patches.
267-283: LGTM!Good handling of the empty string case as
Pendingfor newly created BIOSVersions that haven't been processed yet.
285-313: LGTM!Clean helper functions with proper use of:
clientutils.ListAndFilterControlledByfor ownership-based filteringclient.MergeFromfor conflict-free status patches
350-359: LGTM!Good use of:
Owns(&metalv1alpha1.BIOSVersion{})to trigger reconciliation on owned resource changespredicate.LabelChangedPredicate{}to limit Server watches to label changes, avoiding unnecessary reconciliations on status updates
Proposed Changes
Harmonize and simplify the reconciler code.
Summary by CodeRabbit
✏️ Tip: You can customize this high-level summary in your review settings.