[
https://issues.apache.org/jira/browse/FLINK-4397?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15421233#comment-15421233
]
Stephan Ewen commented on FLINK-4397:
-------------------------------------
Looking at the debug logs, it seems the TaskManagers start the shutdown, but
get stuck in the shutdown at some point after closing the blob cache.
It looks like they get stuck either in
- The network stack disassociation
- The MetricRegistry shutdown.
I wonder if the metric registry has concurrency issues (two concurrent
shutdowns may deadlock?)
> Unstable test SlotCountExceedingParallelismTest.tearDown
> --------------------------------------------------------
>
> Key: FLINK-4397
> URL: https://issues.apache.org/jira/browse/FLINK-4397
> Project: Flink
> Issue Type: Bug
> Reporter: Kostas Kloudas
> Labels: test-stability
>
> An instance can be found here:
> https://s3.amazonaws.com/archive.travis-ci.org/jobs/152392524/log.txt
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
