[
https://issues.apache.org/jira/browse/IGNITE-957?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Gianfranco Murador reassigned IGNITE-957:
-----------------------------------------
Assignee: Gianfranco Murador
> Rest handlers seems to do excessive submissions of the handling logic to
> parallel threads
> -----------------------------------------------------------------------------------------
>
> Key: IGNITE-957
> URL: https://issues.apache.org/jira/browse/IGNITE-957
> Project: Ignite
> Issue Type: Bug
> Components: newbie
> Reporter: Yakov Zhdanov
> Assignee: Gianfranco Murador
>
> Lets trace execution of cache PUT command:
> # request read by HTTP or TCP server
> # then flow gets to
> org.apache.ignite.internal.processors.rest.GridRestProcessor#handleAsync0
> # logic is submitted to REST pool -
> {{org/apache/ignite/internal/processors/rest/GridRestProcessor.java:139}} -
> which is correct since we need to free io (nio) worker thread.
> # (THIS CAN BE REMOVED) then one more submission - system pool submission -
> see {{ctx.closure().callAsync()}} at
> {{org.apache.ignite.internal.processors.rest.handlers.cache.GridCacheCommandHandler#executeCommand(java.util.UUID,
> java.util.UUID, java.lang.String, boolean, java.lang.Object,
> org.apache.ignite.internal.processors.rest.handlers.cache.GridCacheCommandHandler.CacheProjectionCommand)}}
> So, the points are:
> # Flow of each command has to be traced and excessive submissions has to be
> removed
> # Changes should be tested for all protocols: HTTP and MEMCACHED. TCP is less
> prio since it will be dropped once client communication is ready.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
