[
https://issues.apache.org/jira/browse/IGNITE-8809?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16520590#comment-16520590
]
Ivan Daschinskiy commented on IGNITE-8809:
------------------------------------------
First steps are ready. Added and tested special RebalanceRequestMessage and
corresponding logic to ExchangeManager and ExchangeFuture.
Trigger rebalancing by resetting updateCounters to 0 and change state of
partition from owning to moving. Process special case when in message user
request rebalancing for all owners.
> Add ability to control.sh to force rebalance for specific partitions on given
> nodes.
> ------------------------------------------------------------------------------------
>
> Key: IGNITE-8809
> URL: https://issues.apache.org/jira/browse/IGNITE-8809
> Project: Ignite
> Issue Type: Improvement
> Reporter: Alexei Scherbakov
> Assignee: Ivan Daschinskiy
> Priority: Major
> Fix For: 2.6
>
>
> Sometimes it's desirable to force rebalance for specific partitions on given
> nodes, for example, for test reasons or fixing synchronizations issues
> without nodes downtime.
> control.sh should contain new command: rebalance, which will execute the
> exchange request carried by new message type, containing partitions for
> rebalancing and mode: full (evict + move) or delta (historical, using
> counters).
> Example:
> control.sh --rebalance [full|delta] nodeId:p1,p2,p3 node2:p4,p5 ...
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
