[
https://issues.apache.org/jira/browse/FLINK-14337?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16946596#comment-16946596
]
Chesnay Schepler commented on FLINK-14337:
------------------------------------------
The problem is that the archives may not be fully written while the HS is
started. This can lead to exceptions when unpacking them, which shouldn't be a
problem; this easily happens in production and the HS should simply retry on
the next run.
However this retry doesn't happen. The HS adds the job to the list of processed
archives before doing any processing, and only removes the entry if it runs
into an IOException; but this is bypassed since it is an NPE.
The easy fix is to only add the job to the list of processed archives if
everything went smoothly.
> HistoryServerTest.testHistoryServerIntegration failed on Travis
> ---------------------------------------------------------------
>
> Key: FLINK-14337
> URL: https://issues.apache.org/jira/browse/FLINK-14337
> Project: Flink
> Issue Type: Bug
> Components: Runtime / Coordination
> Affects Versions: 1.8.2, 1.9.0, 1.10.0
> Reporter: Till Rohrmann
> Assignee: Chesnay Schepler
> Priority: Critical
> Labels: test-stability
>
> The {{HistoryServerTest.testHistoryServerIntegration}} failed on Travis with
> {code}
> [ERROR] testHistoryServerIntegration[Flink version less than 1.4:
> false](org.apache.flink.runtime.webmonitor.history.HistoryServerTest) Time
> elapsed: 10.667 s <<< FAILURE!
> java.lang.AssertionError: expected:<3> but was:<2>
> {code}
> https://api.travis-ci.org/v3/job/594533358/log.txt
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
