[BUG] BroadcastJob activeDeadlineSeconds did not take effect

[ls-2018]

What happened:

pod is over 300 seconds, but still there

What you expected to happen:

pod cleared

How to reproduce it (as minimally and precisely as possible):

echo 'kind: Cluster
apiVersion: kind.x-k8s.io/v1alpha4
featureGates:
  "EphemeralContainers": true
nodes:
- role: control-plane
  image: kindest/node:v1.26.0
  extraPortMappings:
  - containerPort: 6443
    hostPort: 6443
    protocol: TCP
- role: worker
  image: kindest/node:v1.26.0
  labels:
    zone: c
' >/tmp/kind.yaml

kind create cluster --config /tmp/kind.yaml
kubectl cluster-info --context kind-kind

helm install kruise openkruise/kruise --version 1.4.0


echo 'apiVersion: apps.kruise.io/v1alpha1
kind: BroadcastJob
metadata:
  name: broadcastjob-ttl
spec:
  template:
    spec:
      containers:
        - name: pi
          image: registry.cn-hangzhou.aliyuncs.com/acejilam/tensorflow:latest-gpu     # 5.7G
      restartPolicy: OnFailure
  completionPolicy:
    type: Always
    activeDeadlineSeconds: 300
    ttlSecondsAfterFinished: 300
  paused: false
  parallelism: 3
' >/tmp/bcj.yaml
kubectl apply -f /tmp/bcj.yaml

Anything else we need to know?:

But it doesn't happen every time, and I've found that when the watch pod, node, or BroadcastJob have no events, the tuning logic doesn't fire
 

Environment:

• Kruise version:
• Kubernetes version (use kubectl version):
• Install details (e.g. helm install args):
• Others:

I changed the activeDeadlineSeconds to 30 seconds, and I did it again.

[stale]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.