[
https://issues.apache.org/jira/browse/IGNITE-22838?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17872956#comment-17872956
]
Roman Puchkovskiy commented on IGNITE-22838:
--------------------------------------------
The patch looks good to me
> Operation returns to partition pool after it is handled in RAFT
> ---------------------------------------------------------------
>
> Key: IGNITE-22838
> URL: https://issues.apache.org/jira/browse/IGNITE-22838
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Assignee: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
> Time Spent: 20m
> Remaining Estimate: 0h
>
> h3. Motivation
> After executing the synchronous embedded insert operation, we are returning
> to the partition-operations pool to return control to the client pool. We are
> just losing several microseconds instead of returning to the client pool
> directly.
> {noformat}
> finishFullTx:%node_3344%partition-operations-2 1.3 5476200 5477500
> Here is hidden 5.5 us
> kvPutEndMark:org.apache.ignite.internal.benchmark.InsertBenchmark.kvInsert-jmh-worker-1
> 0.0 5483000 5483000
> {noformat}
> h3. Definition of done
> We need to develop a strategy that does not involve the extra thread in
> synchronous operation.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
