[YARN-8521] NPE in AllocationTagsManager when a container is removed more than once

• Type: Bug
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 3.1.0
• Fix Version/s: 3.2.0, 3.1.2
• Component/s: resourcemanager
• Hadoop Flags: Reviewed
• Assignee: Weiwei Yang
• Reporter: Weiwei Yang
• Created: 12/Jul/18 11:03
• Updated: 10/Aug/18 01:06
• Resolved: 10/Aug/18 00:49

Description

We've seen sometimes there is NPE in AllocationTagsManager

private void removeTagFromInnerMap(Map<String, Long> innerMap, String tag) {
  Long count = innerMap.get(tag);
  if (count > 1) { // NPE!!
  ...

it seems AllocationTagsManager#removeContainer somehow gets called more than once for a same container.