[jira] [Commented] (HBASE-15479) No more garbage or beware of autoboxing

[Vladimir Rodionov (JIRA)]

    [ 
https://issues.apache.org/jira/browse/HBASE-15479?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15200454#comment-15200454
 ] 

Vladimir Rodionov commented on HBASE-15479:
-------------------------------------------

The issue.

AsyncProcess.canTakeOperation before patch:

{code}
 protected boolean canTakeOperation(HRegionLocation loc,
                                     Map<Long, Boolean> regionsIncluded,
                                     Map<ServerName, Boolean> serversIncluded) {
    long regionId = loc.getRegionInfo().getRegionId();
    Boolean regionPrevious = regionsIncluded.get(regionId);

{code}

after patch:

{code}
 protected boolean canTakeOperation(HRegionLocation loc,
                                     Map<Long, Boolean> regionsIncluded,
                                     Map<ServerName, Boolean> serversIncluded) {
    Long regionId = loc.getRegionInfo().getRegionId();
    Boolean regionPrevious = regionsIncluded.get(regionId);

{code}

The difference?  Object allocation on a client side dropped from 120G to 20G in 
the same test run. The test uses BufferedMutator to send data to HBase.




> No more garbage or beware of autoboxing
> ---------------------------------------
>
>                 Key: HBASE-15479
>                 URL: https://issues.apache.org/jira/browse/HBASE-15479
>             Project: HBase
>          Issue Type: Bug
>          Components: Client
>    Affects Versions: 2.0.0
>            Reporter: Vladimir Rodionov
>            Assignee: Vladimir Rodionov
>             Fix For: 2.0.0
>
>         Attachments: HBASE-15479-v1.patch
>
>
> Quick journey with JMC in profile mode revealed very interesting and 
> unexpected heap polluter on a client side. Patch will shortly follow. 



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)