Skip to content
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

fix: failover if master pod is not ready #234

Merged
merged 1 commit into from
Aug 22, 2024
Merged

fix: failover if master pod is not ready #234

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
36 changes: 9 additions & 27 deletions internal/controller/dragonfly_pod_lifecycle_controller.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,23 +75,16 @@ func (r *DfPodLifeCycleReconciler) Reconcile(ctx context.Context, req ctrl.Reque
// Get the role of the pod
role, roleExists := pod.Labels[resources.Role]
if !isPodReady {
restartCount := getRestartCount(pod)
if roleExists && role == "master" {
// If the master Pod is not ready and has restarted atleast once, initiate failover
if restartCount > 0 {
log.Info("Master pod is not starting after multiple attempts, initiating failover", "pod", req.NamespacedName, "restarts", restartCount)
err := dfi.configureReplication(ctx)
if err != nil {
log.Error(err, "Failed to initiate failover")
return ctrl.Result{RequeueAfter: 5 * time.Second}, err
}
return ctrl.Result{}, nil
} else {
log.Info("Master pod is not ready yet, will requeue", "pod", req.NamespacedName, "restarts", restartCount)
return ctrl.Result{RequeueAfter: 5 * time.Second}, nil
log.Info("Master pod is not ready, initiating failover", "pod", req.NamespacedName)
err := dfi.configureReplication(ctx)
if err != nil {
log.Error(err, "Failed to initiate failover")
return ctrl.Result{RequeueAfter: 5 * time.Second}, err
}
return ctrl.Result{}, nil
} else {
log.Info("Pod is not ready yet", "pod", req.NamespacedName, "restarts", restartCount)
log.Info("Pod is not ready yet", "pod", req.NamespacedName)
return ctrl.Result{RequeueAfter: 5 * time.Second}, nil
}
}
Expand All @@ -109,9 +102,9 @@ func (r *DfPodLifeCycleReconciler) Reconcile(ctx context.Context, req ctrl.Reque
// - If the Dragonfly object status is Ready, Then this is a pod restart.
// - If there is master already, add this as replica
// - If there is no master, find a healthy instance and mark it as master
role, ok := pod.Labels[resources.Role]
//
// New pod with No resources.Role
if !ok {
if !roleExists {
log.Info("No replication role was set yet", "phase", dfi.df.Status.Phase)
if dfi.df.Status.Phase == PhaseResourcesCreated {
// Make it ready
Expand Down Expand Up @@ -196,17 +189,6 @@ func (r *DfPodLifeCycleReconciler) Reconcile(ctx context.Context, req ctrl.Reque
return ctrl.Result{}, nil
}

// getRestartCount fetches the restart count for the given dragonfly pod.
func getRestartCount(pod corev1.Pod) int32 {
var restartCount int32 = 0
for _, cs := range pod.Status.ContainerStatuses {
if cs.Name == "dragonfly" {
restartCount += cs.RestartCount
}
}
return restartCount
}

// SetupWithManager sets up the controller with the Manager.
func (r *DfPodLifeCycleReconciler) SetupWithManager(mgr ctrl.Manager) error {
return ctrl.NewControllerManagedBy(mgr).
Expand Down
Loading