[FLINK-10868] Flink's JobCluster ResourceManager doesn't use maximum-failed-containers as limit of resource acquirement

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.6.2, 1.7.0
• Fix Version/s: 1.13.0
• Component/s: Deployment / Mesos, Deployment / YARN
• Labels: pull-request-available
• Assignee: Zhenqiu Huang
• Reporter: Zhenqiu Huang
• Created: 13/Nov/18 17:15
• Updated: 19/Nov/22 09:48
• Resolved: 13/Jan/21 01:09

Description

Currently, YarnResourceManager does use yarn.maximum-failed-containers as limit of resource acquirement. In worse case, when new start containers consistently fail, YarnResourceManager will goes into an infinite resource acquirement process without failing the job. Together with the https://issues.apache.org/jira/browse/FLINK-10848, It will quick occupy all resources of yarn queue.

Issue Links

• Blocked: FLINK-30095 Flink's JobCluster ResourceManager should throw an exception when the failure number of starting worker reaches the maximum failure rate
• causes: FLINK-21139 ThresholdMeterTest.testMarkMultipleEvents unstable
• fixes: FLINK-17127 Make pod creating retry interval configurable
• links to: GitHub Pull Request #8952