hbase-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Hadoop QA (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (HBASE-7874) Allow RegionServer to abort "Put" if it detects that the HBase Client got SocketTimeoutException and disconnected.
Date Tue, 19 Feb 2013 06:41:12 GMT

    [ https://issues.apache.org/jira/browse/HBASE-7874?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13581053#comment-13581053

Hadoop QA commented on HBASE-7874:

{color:red}-1 overall{color}.  Here are the results of testing the latest attachment 
  against trunk revision .

    {color:green}+1 @author{color}.  The patch does not contain any @author tags.

    {color:red}-1 tests included{color}.  The patch doesn't appear to include any new or modified
                        Please justify why no new tests are needed for this patch.
                        Also please list what manual steps were performed to verify this patch.

    {color:red}-1 patch{color}.  The patch command could not apply the patch.

Console output: https://builds.apache.org/job/PreCommit-HBASE-Build/4461//console

This message is automatically generated.
> Allow RegionServer to abort "Put" if it detects that the HBase Client got SocketTimeoutException
and disconnected.
> ------------------------------------------------------------------------------------------------------------------
>                 Key: HBASE-7874
>                 URL: https://issues.apache.org/jira/browse/HBASE-7874
>             Project: HBase
>          Issue Type: Improvement
>          Components: regionserver
>    Affects Versions: 0.94.5
>            Reporter: Maryann Xue
>            Assignee: Maryann Xue
>            Priority: Minor
>         Attachments: HBASE-7874-0.94.patch
> Usually, when regionserver cannot catch up with the put load given by the client, what
happens is region server starts to block update requests from the client until required resource
has been reclaimed (i.e. memstore has been flushed). 
> But in more severe situations, the blocking time gets so long that the client begins
to have SocketTimeoutException and then decides to retry, while in fact the updates are written
into memstore later after they are unblocked. Even though the client has something like a
binary rollback for retry intervals, this can still lead to a vicious circle, leaving the
client to have very low throughput.
> Think we can enable an option to allow regionserver to check if the client has disconnected
(just like what we do in scan) after coming back from "blocking", so that the regionserver
has the same view as the client on whether updates are successfully committed.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message