Skip to content

controller: add precedence-aware setPoolDegraded helper #112

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
jlebon merged 1 commit into from
Jul 16, 2026
Merged

controller: add precedence-aware setPoolDegraded helper #112

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 3 additions & 1 deletion api/v1alpha1/bootcnodepool_types.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -30,7 +30,9 @@ const (
PoolPaused string = "Paused"
)

// PoolDegraded condition reasons.
// PoolDegraded condition reasons. Precedence ordering (which reason
// takes priority when multiple apply) is defined by
// poolDegradedPrecedence in the controller package.
const (
// PoolNodeConflict means a node's labels match multiple pool selectors.
PoolNodeConflict string = "NodeConflict"
Expand Down
74 changes: 45 additions & 29 deletions internal/controller/bootcnodepool_controller.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -327,12 +327,7 @@ func (r *BootcNodePoolReconciler) resolveTargetDigest(ctx context.Context, pool
digest, err := r.TagResolver.Resolve(ctx, pool.Spec.Image.Ref)
if err != nil {
log.Error(err, "Failed to resolve tag", "ref", pool.Spec.Image.Ref)
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: bootcv1alpha1.PoolTagResolutionError,
Message: err.Error(),
})
setPoolDegraded(pool, bootcv1alpha1.PoolTagResolutionError, err.Error())
} else {
if pool.Status.TargetDigest != digest {
log.Info("Resolved tag to new digest", "ref", pool.Spec.Image.Ref, "digest", digest)
Expand Down Expand Up @@ -374,12 +369,7 @@ func isInvalidSpecError(err error) bool {
// setInvalidSpecCondition sets Degraded/InvalidSpec on the pool and
// returns (Result, nil) so Reconcile stops without requeueing.
func (r *BootcNodePoolReconciler) setInvalidSpecCondition(ctx context.Context, pool *bootcv1alpha1.BootcNodePool, specErr error) (ctrl.Result, error) {
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: bootcv1alpha1.PoolInvalidSpec,
Message: specErr.Error(),
})
setPoolDegraded(pool, bootcv1alpha1.PoolInvalidSpec, specErr.Error())
if err := r.Status().Update(ctx, pool); err != nil {
return ctrl.Result{}, fmt.Errorf("updating pool status: %w", err)
}
Expand Down Expand Up @@ -460,11 +450,16 @@ func (r *BootcNodePoolReconciler) syncMembership(ctx context.Context, pool *boot
}

// Set or clear the conflict condition based on what we found.
var conflictingPools []string
for name := range conflicting {
conflictingPools = append(conflictingPools, name)
if len(conflicting) > 0 {
var conflictingPools []string
for name := range conflicting {
conflictingPools = append(conflictingPools, name)
}
// Sort so the message is stable across reconciles.
setPoolDegraded(pool, bootcv1alpha1.PoolNodeConflict,
fmt.Sprintf("Node selector overlaps with pool(s): %s",
strings.Join(slices.Sorted(slices.Values(conflictingPools)), ", ")))
}
syncConflictCondition(pool, conflictingPools)

return ownedSet, nil
}
Expand Down Expand Up @@ -644,18 +639,39 @@ func (r *BootcNodePoolReconciler) restoreCordonState(ctx context.Context, bn *bo
return nil
}

// syncConflictCondition sets or clears the Degraded condition with
// reason NodeConflict on the pool. It only mutates the in-memory
// object; the caller is responsible for writing status.
func syncConflictCondition(pool *bootcv1alpha1.BootcNodePool, conflictingPools []string) {
if len(conflictingPools) > 0 {
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: bootcv1alpha1.PoolNodeConflict,
// Sort so the message is stable across reconciles.
Message: fmt.Sprintf("Node selector overlaps with pool(s): %s",
strings.Join(slices.Sorted(slices.Values(conflictingPools)), ", ")),
})
// poolDegradedPrecedence defines the priority ordering of PoolDegraded
// condition reasons. When multiple degraded reasons apply in a single
// reconcile, the highest-precedence reason wins. setPoolDegraded uses this to
// avoid overwriting a higher-precedence reason with a lower one.
//
// InvalidSpec and TagResolutionError cause early returns in Reconcile and
// don't coexist with the others in practice, but are included for
// completeness.
var poolDegradedPrecedence = map[string]int{
bootcv1alpha1.PoolHealthy: 0,
bootcv1alpha1.PoolNodeDegraded: 1,
bootcv1alpha1.PoolNodeConflict: 2,
bootcv1alpha1.PoolRolloutHalted: 3,
bootcv1alpha1.PoolInvalidSpec: 4,
bootcv1alpha1.PoolTagResolutionError: 5,
}

// setPoolDegraded sets the Degraded condition on the pool, but only if the new
// reason has higher precedence than the current one. This prevents e.g.
// RolloutHalted from overwriting NodeConflict. Unknown reasons get precedence
// 0 (map zero-value), so they silently no-op. Keep poolDegradedPrecedence in
// sync with the API constants.
func setPoolDegraded(pool *bootcv1alpha1.BootcNodePool, reason, message string) {

@alicefr alicefr Jul 16, 2026

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We don't have unit tests for such helpers functions yet, but it might be helpful to add some. The function is relatively straightforward so up to you. But it might be good to introduce it as a good habit

@jlebon jlebon Jul 16, 2026

Copy link
Copy Markdown
Collaborator Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

We do have a few function-level unit tests though I skipped adding one for this one because it just didn't seem worth it for the complexity.

current := apimeta.FindStatusCondition(pool.Status.Conditions, bootcv1alpha1.PoolDegraded)
if current != nil && current.Status == metav1.ConditionTrue {
if poolDegradedPrecedence[reason] <= poolDegradedPrecedence[current.Reason] {
return
}
}
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: reason,
Message: message,
})
}
60 changes: 14 additions & 46 deletions internal/controller/rollout.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -68,10 +68,13 @@ func (r *BootcNodePoolReconciler) driveRollout(ctx context.Context, pool *bootcv

rs := buildRolloutState(log, ownedBootcNodes)

// Flag degraded nodes at the pool level. NodeConflict (set during
// membership sync) takes priority, so we only set NodeDegraded if
// the pool isn't already degraded for another reason.
syncNodeDegradedCondition(pool, rs.degraded)
// Flag degraded nodes at the pool level.
if len(rs.degraded) > 0 {
names := nodeNames(rs.degraded)
slices.Sort(names)
setPoolDegraded(pool, bootcv1alpha1.PoolNodeDegraded,
fmt.Sprintf("Degraded nodes: %s", strings.Join(names, ", ")))
}

// Free reboot slots for nodes that have successfully rebooted into
// the desired image. This runs before computing available slots so
Expand All @@ -88,7 +91,13 @@ func (r *BootcNodePoolReconciler) driveRollout(ctx context.Context, pool *bootcv

rolloutHalted := len(unhealthy) >= unhealthySlotHaltThreshold
if rolloutHalted {
syncRolloutHaltedCondition(pool, unhealthy)
details := make([]string, len(unhealthy))
for i, u := range unhealthy {
details[i] = u.name + ": " + u.reason
}
slices.Sort(details)
setPoolDegraded(pool, bootcv1alpha1.PoolRolloutHalted,
fmt.Sprintf("Rollout halted: 2+ unhealthy nodes in reboot slots (%s)", strings.Join(details, ", ")))
log.Info("Rollout halted: 2+ unhealthy nodes in reboot slots",
"unhealthyInSlots", len(unhealthy))

Expand Down Expand Up @@ -393,29 +402,6 @@ func buildRolloutState(log logr.Logger, ownedBootcNodes map[string]*bootcv1alpha
return rs
}

// syncNodeDegradedCondition sets Degraded/NodeDegraded on the pool if any
// nodes are degraded. It respects priority: if the pool is already degraded
// for another reason (e.g. NodeConflict), it does not overwrite it.
// XXX(jl): Or... should this be a new condition type so it can be surfaced in
// parallel? Let's see how this approach feels and iterate.
func syncNodeDegradedCondition(pool *bootcv1alpha1.BootcNodePool, degraded []*bootcv1alpha1.BootcNode) {
if len(degraded) == 0 {
return
}
if apimeta.IsStatusConditionTrue(pool.Status.Conditions, bootcv1alpha1.PoolDegraded) {
return
}
names := nodeNames(degraded)
// Sort so the message is stable across reconciles.
slices.Sort(names)
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: bootcv1alpha1.PoolNodeDegraded,
Message: fmt.Sprintf("Degraded nodes: %s", strings.Join(names, ", ")),
})
}

// unhealthySlot identifies a node in a reboot slot that is unhealthy.
type unhealthySlot struct {
name string
Expand Down Expand Up @@ -452,24 +438,6 @@ func findUnhealthySlots(rs *rolloutState, targetDigest string) []unhealthySlot {
return result
}

// syncRolloutHaltedCondition sets Degraded/RolloutHalted on the pool. It
// implicitly takes priority over NodeDegraded (which checks for existing
// daemon-driven degraded status) by running later than
// syncNodeDegradedCondition.
func syncRolloutHaltedCondition(pool *bootcv1alpha1.BootcNodePool, unhealthy []unhealthySlot) {
details := make([]string, len(unhealthy))
for i, u := range unhealthy {
details[i] = u.name + ": " + u.reason
}
slices.Sort(details)
apimeta.SetStatusCondition(&pool.Status.Conditions, metav1.Condition{
Type: bootcv1alpha1.PoolDegraded,
Status: metav1.ConditionTrue,
Reason: bootcv1alpha1.PoolRolloutHalted,
Message: fmt.Sprintf("Rollout halted: 2+ unhealthy nodes in reboot slots (%s)", strings.Join(details, ", ")),
})
}

// resolveMaxUnavailable computes the effective maxUnavailable value from the
// pool's rollout spec. Defaults to 1 when unset. A value of 0 is allowed and
// means no reboot slots are available (effectively paused). Returns an
Expand Down