[
https://issues.apache.org/jira/browse/IGNITE-1093?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14947196#comment-14947196
]
Anton Vinogradov commented on IGNITE-1093:
------------------------------------------
I've implemented new version of rebalancing, testing and benchmarking now.
Found really weird thing:
I see that in case when new node join topology at the end of rebalancing at
supply node happens big GC pause (not stop the world).
249.920: [GC249.920: [ParNew: 130944K->0K(131008K), 0.3429600 secs]
6010351K->5945518K(15728576K), 0.3431210 secs] [Times: user=1.98 sys=0.01,
real=0.35 secs]
In case failureDetectionTimeout less than 300 ms this node left topology.
Same happens at both new and old rebalancing implementation.
Going to make heapdumps at the end of rebalancing to fix this issue.
In case someone know possible reasons please feel free to post ideas.
> Rebalancing with default parameters is very slow
> ------------------------------------------------
>
> Key: IGNITE-1093
> URL: https://issues.apache.org/jira/browse/IGNITE-1093
> Project: Ignite
> Issue Type: Bug
> Components: cache
> Affects Versions: sprint-7
> Reporter: Pavel Konstantinov
> Assignee: Anton Vinogradov
> Priority: Critical
> Fix For: ignite-1.5
>
> Attachments: Plot_ThroughputLatencyProbe_01.png, rebalancing.zip
>
>
> # Start one node with partitioned cache with one backup.
> # Load into the cache 40billions of keys using DataStreamer
> # Start second node on the same host
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
