You are viewing a plain text version of this content. The canonical link for it is here.
Posted to issues@ignite.apache.org by "Stanilovsky Evgeny (Jira)" <ji...@apache.org> on 2020/02/28 05:29:00 UTC
[jira] [Commented] (IGNITE-10720) Decrease time to save metadata
during checkpoint
[ https://issues.apache.org/jira/browse/IGNITE-10720?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17047217#comment-17047217 ]
Stanilovsky Evgeny commented on IGNITE-10720:
---------------------------------------------
guys looks like there is deadlock, check :
{noformat}fillCacheGroupState{noformat} already under WLock, and a little bit near we wants RLock:
{noformat}part.dataStore().fullSize(){noformat} proofs here [1]
[1] https://ci.ignite.apache.org/viewType.html?buildTypeId=IgniteTests24Java8_PdsIndexing&tab=buildTypeStatusDiv&branch_IgniteTests24Java8=<default>
> Decrease time to save metadata during checkpoint
> ------------------------------------------------
>
> Key: IGNITE-10720
> URL: https://issues.apache.org/jira/browse/IGNITE-10720
> Project: Ignite
> Issue Type: Improvement
> Reporter: Anton Kalashnikov
> Assignee: Anton Kalashnikov
> Priority: Major
> Fix For: 2.8
>
> Time Spent: 40m
> Remaining Estimate: 0h
>
> Looks like it's not neccessery save all metadata(like free list) under write checkpoint lock because sometimes it's too long.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)