You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@hbase.apache.org by Ted Yu <yu...@gmail.com> on 2012/01/18 19:00:54 UTC
code review: HBASE-5179 Concurrent processing of processFaileOver and
ServerShutdownHandler may cause region to be assigned before log splitting is
completed, causing data loss
Hi,
There have been multiple iterations for HBASE-5179.
Please review 5179-v11.txt and provide your comments. Some of your initial
questions should have been answered by the long discussion on the JIRA.
Cheers
Re: code review: HBASE-5179 Concurrent processing of processFaileOver
and ServerShutdownHandler may cause region to be assigned before log
splitting is completed, causing data loss
Posted by Ted Yu <yu...@gmail.com>.
https://issues.apache.org/jira/browse/HBASE-4748 provides the scenario
which 5179-90v17.txt fixes.
FYI
On Fri, Jan 20, 2012 at 1:50 PM, Ted Yu <yu...@gmail.com> wrote:
> Writing unit test is a good idea.
>
> See the following comments for scenarios:
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186099&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186099
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186208&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186208
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186720&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186720
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13188979&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13188979
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13188983&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13188983
>
> https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13189186&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13189186
>
> Cheers
>
>
> On Fri, Jan 20, 2012 at 1:31 PM, Stack <st...@duboce.net> wrote:
>
>> On Fri, Jan 20, 2012 at 9:57 AM, Ted Yu <yu...@gmail.com> wrote:
>>
>> > There have been several rounds of bug fixing since v11.
>> >
>> > Please take a look at 5179-90v17.txt which passed test on Jinchao's
>> > cluster.
>> > Once code review passed for 5179-90v17.txt, I will prepare patches for
>> > TRUNK and 0.92 branch.
>> >
>> >
>> In review of v11, I suggested we needed tests or at least test scenarios
>> outlined (if latter done I volunteered to try and write tests for trunk at
>> least) especially because I'm unclear on what we're actually fixing with
>> the patch at this stage. I also suggested breaking up the patch into
>> smaller, targeted pieces as a means of undoing the complexity so mere
>> mortals like myself can follow along. I don't see this done in v17.
>>
>> St.Ack
>>
>
>
Re: code review: HBASE-5179 Concurrent processing of processFaileOver
and ServerShutdownHandler may cause region to be assigned before log
splitting is completed, causing data loss
Posted by Ted Yu <yu...@gmail.com>.
Writing unit test is a good idea.
See the following comments for scenarios:
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186099&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186099
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186208&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186208
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13186720&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13186720
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13188979&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13188979
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13188983&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13188983
https://issues.apache.org/jira/browse/HBASE-5179?focusedCommentId=13189186&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-13189186
Cheers
On Fri, Jan 20, 2012 at 1:31 PM, Stack <st...@duboce.net> wrote:
> On Fri, Jan 20, 2012 at 9:57 AM, Ted Yu <yu...@gmail.com> wrote:
>
> > There have been several rounds of bug fixing since v11.
> >
> > Please take a look at 5179-90v17.txt which passed test on Jinchao's
> > cluster.
> > Once code review passed for 5179-90v17.txt, I will prepare patches for
> > TRUNK and 0.92 branch.
> >
> >
> In review of v11, I suggested we needed tests or at least test scenarios
> outlined (if latter done I volunteered to try and write tests for trunk at
> least) especially because I'm unclear on what we're actually fixing with
> the patch at this stage. I also suggested breaking up the patch into
> smaller, targeted pieces as a means of undoing the complexity so mere
> mortals like myself can follow along. I don't see this done in v17.
>
> St.Ack
>
Re: code review: HBASE-5179 Concurrent processing of processFaileOver
and ServerShutdownHandler may cause region to be assigned before log
splitting is completed, causing data loss
Posted by Stack <st...@duboce.net>.
On Fri, Jan 20, 2012 at 9:57 AM, Ted Yu <yu...@gmail.com> wrote:
> There have been several rounds of bug fixing since v11.
>
> Please take a look at 5179-90v17.txt which passed test on Jinchao's
> cluster.
> Once code review passed for 5179-90v17.txt, I will prepare patches for
> TRUNK and 0.92 branch.
>
>
In review of v11, I suggested we needed tests or at least test scenarios
outlined (if latter done I volunteered to try and write tests for trunk at
least) especially because I'm unclear on what we're actually fixing with
the patch at this stage. I also suggested breaking up the patch into
smaller, targeted pieces as a means of undoing the complexity so mere
mortals like myself can follow along. I don't see this done in v17.
St.Ack
Re: code review: HBASE-5179 Concurrent processing of processFaileOver
and ServerShutdownHandler may cause region to be assigned before log
splitting is completed, causing data loss
Posted by Ted Yu <yu...@gmail.com>.
There have been several rounds of bug fixing since v11.
Please take a look at 5179-90v17.txt which passed test on Jinchao's cluster.
Once code review passed for 5179-90v17.txt, I will prepare patches for
TRUNK and 0.92 branch.
Cheers
On Wed, Jan 18, 2012 at 10:00 AM, Ted Yu <yu...@gmail.com> wrote:
> Hi,
> There have been multiple iterations for HBASE-5179.
>
> Please review 5179-v11.txt and provide your comments. Some of your initial
> questions should have been answered by the long discussion on the JIRA.
>
> Cheers
>