You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@airflow.apache.org by "Zill Silveira (Jira)" <ji...@apache.org> on 2021/07/20 09:21:00 UTC
[jira] [Commented] (AIRFLOW-6806) Kubernetes Executor : Sometimes,
task is succeeded, but pod is not completed ( keep running ) . Zombie pod
is born.
[ https://issues.apache.org/jira/browse/AIRFLOW-6806?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17383929#comment-17383929 ]
Zill Silveira commented on AIRFLOW-6806:
----------------------------------------
[~Yuya] I am getting the same Zombie POD issue as you have mentioned. The Task Pod completes it's execution and the process goes to zombie state due to which the POD keeps running. How did you fix the zombie pods.
> Kubernetes Executor : Sometimes, task is succeeded, but pod is not completed ( keep running ) . Zombie pod is born.
> --------------------------------------------------------------------------------------------------------------------
>
> Key: AIRFLOW-6806
> URL: https://issues.apache.org/jira/browse/AIRFLOW-6806
> Project: Apache Airflow
> Issue Type: Bug
> Components: executor-kubernetes, executors
> Affects Versions: 1.10.9
> Reporter: Matsubara Yuya
> Assignee: Daniel Imberman
> Priority: Blocker
>
> The task completes successfully, and the result is able to check on the UI. However, there are cases where the Pod that was processing a completed task is keep running and will not end forever. A zombie pod is born.
>
> Use Operator like this,
> default_pod = BashOperator(
> task_id="test",
> env=KubePodEnv,
> retries=1,
> xcom_push=True,
> bash_command="python3 /opt/test.py ",
> executor_config={
> "KubernetesExecutor":
>
> { "volumes": volumes, "volume_mounts": volume_mounts, "affinity":affinity, }
> },
> dag=dag
> )
--
This message was sent by Atlassian Jira
(v8.3.4#803005)