You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Alexey Goncharuk (JIRA)" <ji...@apache.org> on 2018/09/21 08:58:00 UTC
[jira] [Comment Edited] (IGNITE-9492) Refactor processing
SingleMessage with exchangeId==null
[ https://issues.apache.org/jira/browse/IGNITE-9492?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16623285#comment-16623285 ]
Alexey Goncharuk edited comment on IGNITE-9492 at 9/21/18 8:57 AM:
-------------------------------------------------------------------
[~Jokser], I see these two test started to consistently fail in the submitted PR:
{code}
WalCompactionTest.testApplyingUpdatesFromCompactedWal WalCompactionTest.testApplyingUpdatesFromCompactedWalWhenCompressorDisabled
{code}
Can you take a look?
https://ci.ignite.apache.org/viewLog.html?buildId=1922084&tab=buildResultsDiv&buildTypeId=IgniteTests24Java8_PdsDirectIo2
was (Author: agoncharuk):
[~Jokser], I see these two test started to consistently fail in the submitted PR:
{code}
WalCompactionTest.testApplyingUpdatesFromCompactedWal WalCompactionTest.testApplyingUpdatesFromCompactedWalWhenCompressorDisabled
{code}
Can you take a look?
> Refactor processing SingleMessage with exchangeId==null
> -------------------------------------------------------
>
> Key: IGNITE-9492
> URL: https://issues.apache.org/jira/browse/IGNITE-9492
> Project: Ignite
> Issue Type: Improvement
> Components: cache
> Affects Versions: 2.5
> Reporter: Pavel Kovalenko
> Assignee: Pavel Kovalenko
> Priority: Major
> Fix For: 2.7
>
>
> Currently, after each PME coordinator spend a lot of time on processing correcting Single messages (with exchange id == null). This leads to growing inbound/outbound messages queue and delaying other coordinator-aware processes.
> Processing single messages with exchange id == null are not so important to give all available resources to it. We should limit the number of sys-threads which are able to process such single messages.
> We shouldn't create a big SingleMessages for each of partition state change and use more shrinked/tiny messages for that.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)