hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sean Busbey (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-15163) Add sampling code and metrics for get/scan/multi/mutate count separately
Date Wed, 27 Jan 2016 05:17:40 GMT

    [ https://issues.apache.org/jira/browse/HBASE-15163?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15118682#comment-15118682
] 

Sean Busbey commented on HBASE-15163:
-------------------------------------

{quote}
bq. org.apache.hadoop.hbase.ipc.RpcServer$Connection.close() might ignore java.lang.Exception
is not caused by this patch.
{quote}

indeed. it is listed as a pre-existing condition:

{quote}
-1	findbugs	2m 12s	hbase-server in master has 1 extant Findbugs warnings.
{quote}

turning this -1 into a 0 is already filed as HBASE-15151

{quote}
The mvninstall issue:
bq. cannot find symbol RPC_MUTATE_REQUEST_COUNT
seems to be a yetus bug or config issue. This variable is a newly added one and included in
the patch, but in another module rather than the currently checking one. Sean Busbey could
you help take a look here? Thanks.
{quote}

this is YETUS-280. you can build with this change in place if you [go to the precommit job|https://builds.apache.org/view/PreCommit%20Builds/job/PreCommit-HBASE-Build/]
and use "build with parameters" to check the USE_YETUS_PRERELEASE option.

{quote}
The checkstyle issue:
bq. RSRpcServices.java:2432:3: error: Method length is 414 lines (max allowed is 150)
has nothing to do with changes here.
{quote}

that's not entirely accurate.  This part of the patch is causing this to be flagged as related
to the patch:

{quote}
@@ -2451,6 +2467,7 @@ public class RSRpcServices implements HBaseRPCErrorHandler,
         throw e;
       }
       requestCount.increment();
+      rpcScanRequestCount.increment();
 
       int ttl = 0;
       Region region = null;
{quote}

It's true that the method {{public ScanResponse scan(final RpcController controller, final
ScanRequest request)}} was already flagged as too large. However this patch is contributing
to making the problem worse by taking the method from 413 lines to 414 lines. I don't have
strong feelings one way or another on wether that needs to be dealt with in this patch.

> Add sampling code and metrics for get/scan/multi/mutate count separately
> ------------------------------------------------------------------------
>
>                 Key: HBASE-15163
>                 URL: https://issues.apache.org/jira/browse/HBASE-15163
>             Project: HBase
>          Issue Type: Sub-task
>            Reporter: Yu Li
>            Assignee: Yu Li
>         Attachments: DifferentRequestQPS.png, HBASE-15163.patch, HBASE-15163_v2.patch,
HBASE-15163_v3.patch, HBASE-15163_v3.patch
>
>
> This way we could see QPS of different kinds of requests, to better analyze what's causing
hot spot in system



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

Mime
View raw message