Helen Weng created YARN-11636: --------------------------------- Summary: App stuck in ACCEPTED state, however Yarn metric thinks there are no pending apps in the queue Key: YARN-11636 URL: https://issues.apache.org/jira/browse/YARN-11636 Project: Hadoop YARN Issue Type: Bug Affects Versions: 3.2.1 Reporter: Helen Weng
Hi, I've encountered a case recently when an app gets stuck in ACCEPTED state forever in a queue. The queue is busy for about 4 hrs so during this time, being stuck in ACCEPTED is expected. However even as resources become available and all other jobs run, this job continues to be stuck. I've checked the following states: 1. Resources are available at the leaf queue and cluster level. 2. Other jobs can get the resources to run 3. Not hitting maxAM limits (there is only 1 other job running during this time in the queue and it is using near 0 resources) 4. When I look at jmx metric it seems to think the app is running. AppsRunning says 1 and containersRunning says 1 while while AppsPending says 0. However the app is staunchly in the "Accepted" state and does not seem to be running. Is this known or have others encountered this issue before? Or do you have any advice on what I can look into to debug it? Thanks very much for the help. -- This message was sent by Atlassian Jira (v8.20.10#820010) --------------------------------------------------------------------- To unsubscribe, e-mail: yarn-dev-unsubscr...@hadoop.apache.org For additional commands, e-mail: yarn-dev-h...@hadoop.apache.org