[jira] [Commented] (IGNITE-5521) Large near caches lead to cluster instability with metrics enabled

["ASF GitHub Bot (JIRA)" <ji...@apache.org>]

    [ https://issues.apache.org/jira/browse/IGNITE-5521?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16064531#comment-16064531 ] 

ASF GitHub Bot commented on IGNITE-5521:
----------------------------------------

GitHub user mcherkasov opened a pull request:

    https://github.com/apache/ignite/pull/2201

    IGNITE-5521: Large near caches lead to cluster instability with metrics enabled

    

You can merge this pull request into a Git repository by running:

    $ git pull https://github.com/gridgain/apache-ignite ignite-5521

Alternatively you can review and apply these changes as the patch at:

    https://github.com/apache/ignite/pull/2201.patch

To close this pull request, make a commit to your master/trunk branch
with (at least) the following in the commit message:

    This closes #2201
    
----

----


> Large near caches lead to cluster instability with metrics enabled
> ------------------------------------------------------------------
>
>                 Key: IGNITE-5521
>                 URL: https://issues.apache.org/jira/browse/IGNITE-5521
>             Project: Ignite
>          Issue Type: Bug
>          Components: cache, general
>    Affects Versions: 1.7
>            Reporter: Alexey Goncharuk
>            Assignee: Mikhail Cherkasov
>            Priority: Critical
>              Labels: important
>             Fix For: 2.1
>
>
> We have two issues in the way cache metrics are working:
> 1) Near cache size is calculated using full iteration over the near entries. Perhaps, this is done because of near entries may be invalidated by a primary node change, however, we should give a less strict metric rather than O(N) cache size time
> 2) Cache metrics are copied in discovery worker threads. This looks a bit risky because an error like the one described before may stall the whole cluster. We need to make sure that when the heartbeat message is processed, we already have a metrics snapshot enabled



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)