[ https://issues.apache.org/jira/browse/YARN-3701?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14556516#comment-14556516 ]
Zhijie Shen commented on YARN-3701: ----------------------------------- bq. Can we do the same thing for getAppAttempts and getContainers ? That's a good question. It depends on the isolation granularity. My opinion is that we set the isolation at the app level, such that app1's error doesn't affect that I access app2, app3 and so on. But anything error inside app1, such as corruption of one container and one attempt, should let user know. Thoughts? > Isolating the error of generating a single app report when getting all apps > from generic history service > -------------------------------------------------------------------------------------------------------- > > Key: YARN-3701 > URL: https://issues.apache.org/jira/browse/YARN-3701 > Project: Hadoop YARN > Issue Type: Bug > Components: timelineserver > Reporter: Zhijie Shen > Assignee: Zhijie Shen > Priority: Blocker > Attachments: YARN-3701.1.patch > > > Nowadays, if some error of generating a single app report when getting the > application list from generic history service, it will throw the exception. > Therefore, even if it just 1 out of 100 apps has something wrong, the whole > app list is screwed. The worst impact is making the default page (app list) > of GHS web UI crash, wile REST API /applicationhistory/apps will also break. -- This message was sent by Atlassian JIRA (v6.3.4#6332)