[jira] [Comment Edited] (GEODE-8189) gfsh shutdown commands generate noisy DistributedSystemDisconnectedException

["Kirk Lund (Jira)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/GEODE-8189?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17116965#comment-17116965 ] 

Kirk Lund edited comment on GEODE-8189 at 5/26/20, 6:35 PM:
------------------------------------------------------------

I'm submitting a PR to fix the flakiness in GEODE-6070 by using IgnoredException. If ShutdownFunction or any other product classes need further changes, please track that here in GEODE-8189.


was (Author: klund):
I'm submitting a PR to fix the flakiness in GEODE-6070 by using IgnoredException. If ShutdownFunction or any other product classes need further changes, please track that in GEODE-8189.

> gfsh shutdown commands generate noisy DistributedSystemDisconnectedException
> ----------------------------------------------------------------------------
>
>                 Key: GEODE-8189
>                 URL: https://issues.apache.org/jira/browse/GEODE-8189
>             Project: Geode
>          Issue Type: Bug
>          Components: gfsh
>            Reporter: Kirk Lund
>            Assignee: Dan Smith
>            Priority: Major
>
> The gfsh shutdown commands generate noisy DistributedSystemDisconnectedException stack traces in the geode log files. This results in flaky failures such as GEODE-6070 but also creates undesired log noise and alerts, etc.
> Dan Smith's analysis from GEODE-6070:
> {quote}Looking at the code, this look like a race condition in ShutdownFunction.execute. It spins up a thread to shutdown the cache, and then sends a reply (using geode's messaging). So maybe the shutdown can happen before the reply? This class then does a blanket catch and log warning. Maybe ignoring CancelException would be ok, or maybe this should somehow reply before shutdown?{quote}



--
This message was sent by Atlassian Jira
(v8.3.4#803005)