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

Keep the populator pod on failure #637

Merged
merged 4 commits into from
Nov 12, 2023
+66 −16
Merged

Keep the populator pod on failure #637

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
20d68b7
Keep the populator pod on failure
liranr23 Oct 31, 2023
4fabbf5
Decreace updatePopulatorCopyProgress complexity
liranr23 Oct 31, 2023
92195e1
Refactor the code
liranr23 Nov 1, 2023
51d4ce3
Save annotation of restarts
liranr23 Nov 7, 2023
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
26 changes: 24 additions & 2 deletions pkg/controller/plan/migration.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1453,7 +1453,7 @@ func (r *Migration) updateCopyProgress(vm *plan.VMStatus, step *plan.Step) (err
if r.Plan.Spec.Warm && len(importer.Status.ContainerStatuses) > 0 {
vm.Warm.Failures = int(importer.Status.ContainerStatuses[0].RestartCount)
}
if RestartLimitExceeded(importer) {
if restartLimitExceeded(importer) {
task.MarkedCompleted()
msg, _ := terminationMessage(importer)
task.AddError(msg)
Expand Down Expand Up @@ -1598,6 +1598,10 @@ func (r *Migration) updatePopulatorCopyProgress(vm *plan.VMStatus, step *plan.St
if err != nil {
return
}
populatorPods, err := r.kubevirt.getPopulatorPods()
if err != nil {
return
}

for _, pvc := range pvcs {
if _, ok := pvc.Annotations["lun"]; ok {
Expand All @@ -1616,6 +1620,11 @@ func (r *Migration) updatePopulatorCopyProgress(vm *plan.VMStatus, step *plan.St
continue
}

err = r.isPopulatorFailed(populatorPods, string(pvc.UID))
if err != nil {
return
}

if pvc.Status.Phase == core.ClaimBound {
task.Phase = Completed
task.Reason = TransferCompleted
Expand All @@ -1638,6 +1647,19 @@ func (r *Migration) updatePopulatorCopyProgress(vm *plan.VMStatus, step *plan.St
return
}

func (r *Migration) isPopulatorFailed(populatorPods []core.Pod, givenPvcId string) error {
liranr23 marked this conversation as resolved.
Show resolved Hide resolved
for _, pod := range populatorPods {
pvcId := strings.Split(pod.Name, "populate-")[1]
liranr23 marked this conversation as resolved.
Show resolved Hide resolved
if givenPvcId != pvcId {
continue
}
if pod.Status.Phase == core.PodFailed {
return fmt.Errorf("populator pod %s/%s failed for PVC %s. Please check the pod logs.", pod.Namespace, pod.Name, pvcId)
}
liranr23 marked this conversation as resolved.
Show resolved Hide resolved
}
return nil
}

func (r *Migration) setPopulatorPodsWithLabels(vm *plan.VMStatus, migrationID string) {
podList, err := r.kubevirt.GetPodsWithLabels(map[string]string{})
if err != nil {
Expand Down Expand Up @@ -1701,7 +1723,7 @@ func terminationMessage(pod *core.Pod) (msg string, ok bool) {
}

// Return whether the pod has failed and restarted too many times.
func RestartLimitExceeded(pod *core.Pod) (exceeded bool) {
func restartLimitExceeded(pod *core.Pod) (exceeded bool) {
if len(pod.Status.ContainerStatuses) == 0 {
return
}
Expand Down
1 change: 0 additions & 1 deletion pkg/lib-volume-populator/populator-machinery/BUILD.bazel
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -10,7 +10,6 @@ go_library(
visibility = ["//visibility:public"],
deps = [
"//pkg/apis/forklift/v1beta1",
"//pkg/controller/plan",
"//vendor/k8s.io/api/core/v1:core",
"//vendor/k8s.io/api/storage/v1:storage",
"//vendor/k8s.io/apimachinery/pkg/api/errors",
Expand Down
8 changes: 0 additions & 8 deletions pkg/lib-volume-populator/populator-machinery/controller.go
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,7 +31,6 @@ import (
"time"

"github.com/konveyor/forklift-controller/pkg/apis/forklift/v1beta1"
"github.com/konveyor/forklift-controller/pkg/controller/plan"
corev1 "k8s.io/api/core/v1"
storagev1 "k8s.io/api/storage/v1"
"k8s.io/apimachinery/pkg/api/errors"
Expand Down Expand Up @@ -698,13 +697,6 @@ func (c *controller) syncPvc(ctx context.Context, key, pvcNamespace, pvcName str
if corev1.PodSucceeded != pod.Status.Phase {
if corev1.PodFailed == pod.Status.Phase {
c.recorder.Eventf(pvc, corev1.EventTypeWarning, reasonPodFailed, "Populator failed: %s", pod.Status.Message)
liranr23 marked this conversation as resolved.
Show resolved Hide resolved
// Delete failed pods so we can try again
if !plan.RestartLimitExceeded(pod) {
err = c.kubeClient.CoreV1().Pods(populatorNamespace).Delete(ctx, pod.Name, metav1.DeleteOptions{})
if err != nil {
return err
}
}
ahadas marked this conversation as resolved.
Show resolved Hide resolved
}
// We'll get called again later when the pod succeeds
return nil
Expand Down
Loading