Fix job recovery and savepoint bug (#401)

elanv · web-flow · commit 7c37f850cef4 · 2021-02-08T21:51:19.000-08:00
diff --git a/controllers/flinkcluster_reconciler.go b/controllers/flinkcluster_reconciler.go
@@ -521,7 +521,7 @@ func (reconciler *ClusterReconciler) reconcileJob() (ctrl.Result, error) {
 			shouldTakeSavepont, savepointTriggerReason := reconciler.shouldTakeSavepoint()
 			if shouldTakeSavepont {
 				err = reconciler.updateSavepointTriggerTimeStatus()
-				if err != nil {
+				if err == nil {
 					newSavepointStatus, _ = reconciler.takeSavepointAsync(jobID, savepointTriggerReason)
 				}
 			}
@@ -997,6 +997,7 @@ func (reconciler *ClusterReconciler) updateStatusForNewJob() error {
 		clusterClone.Status.Components.Job = newJobStatus
 	}
 	var fromSavepoint = getFromSavepoint(desiredJob.Spec)
+	newJobStatus.ID = ""
 	newJobStatus.State = v1beta1.JobStatePending
 	newJobStatus.FromSavepoint = fromSavepoint
 	if newJobStatus.SavepointLocation != "" {

Original file line number	Diff line number	Diff line change
`@@ -521,7 +521,7 @@ func (reconciler *ClusterReconciler) reconcileJob() (ctrl.Result, error) {`
`521`	`521`	`shouldTakeSavepont, savepointTriggerReason := reconciler.shouldTakeSavepoint()`
`522`	`522`	`if shouldTakeSavepont {`
`523`	`523`	`err = reconciler.updateSavepointTriggerTimeStatus()`
`524`		`- if err != nil {`
	`524`	`+ if err == nil {`
`525`	`525`	`newSavepointStatus, _ = reconciler.takeSavepointAsync(jobID, savepointTriggerReason)`
`526`	`526`	`}`
`527`	`527`	`}`
`@@ -997,6 +997,7 @@ func (reconciler *ClusterReconciler) updateStatusForNewJob() error {`
`997`	`997`	`clusterClone.Status.Components.Job = newJobStatus`
`998`	`998`	`}`
`999`	`999`	`var fromSavepoint = getFromSavepoint(desiredJob.Spec)`
	`1000`	`+ newJobStatus.ID = ""`
`1000`	`1001`	`newJobStatus.State = v1beta1.JobStatePending`
`1001`	`1002`	`newJobStatus.FromSavepoint = fromSavepoint`
`1002`	`1003`	`if newJobStatus.SavepointLocation != "" {`