fix(controller): Handling panic in leaderelection (argoproj#6072)

Signed-off-by: Saravanan Balasubramanian <[email protected]>

workflow/controller/controller.go

@@ -192,6 +192,10 @@ var indexers = cache.Indexers{
 // Run starts an Workflow resource controller
 func (wfc *WorkflowController) Run(ctx context.Context, wfWorkers, workflowTTLWorkers, podWorkers, podCleanupWorkers int) {
  defer runtimeutil.HandleCrash(runtimeutil.PanicHandlers...)
+
+ ctx, cancel := context.WithCancel(ctx)
+ defer cancel()
+
  defer wfc.wfQueue.ShutDown()
  defer wfc.podQueue.ShutDown()
  defer wfc.podCleanupQueue.ShutDown()
@@ -240,7 +244,6 @@ func (wfc *WorkflowController) Run(ctx context.Context, wfWorkers, workflowTTLWo
  }
  logCtx := log.WithField("id", nodeID)
 
- var cancel context.CancelFunc
  leaderName := "workflow-controller"
  if wfc.Config.InstanceID != "" {
  leaderName = fmt.Sprintf("%s-%s", leaderName, wfc.Config.InstanceID)