Home | About | Sematext search-lucene.com search-hadoop.com
 Search Hadoop and all its subprojects:

Switch to Threaded View
HBase >> mail # dev >> [Discuss] Release 0.92.2?


Copy link to this message
-
Re: [Discuss] Release 0.92.2?
Thanks for the response Dave.

I exchanged a few emails with Lars H last night and we felt we shouldn't
block 0.92.2 / 0.94.2 releases due to a few flaky tests.

Let's fix flaky tests continuously.

On Thu, Aug 30, 2012 at 3:14 PM, Dave Wang <[EMAIL PROTECTED]> wrote:

> +1 on limiting changes into 0.92.  Just wanted to point out a patch that we
> may be able to pull part of to fix the test.
>
> - Dave
>
> On Thu, Aug 30, 2012 at 2:07 PM, Ted Yu <[EMAIL PROTECTED]> wrote:
>
> > Looking at 5992.v11.patch, it is quite big, touching master,
> > AssignmentManager, etc.
> >
> > At this moment, I want to limit the scope of changes going into 0.92
> > branch.
> >
> > Cheers
> >
> > On Thu, Aug 30, 2012 at 1:55 PM, Dave Wang <[EMAIL PROTECTED]> wrote:
> >
> > > Does HBASE-5992 fix this?  That fix is only in trunk right now.
> > >
> > > - Dave
> > >
> > > On Thu, Aug 30, 2012 at 1:52 PM, Ted Yu <[EMAIL PROTECTED]> wrote:
> > >
> > > > Looking at TestDrainingServer failure:
> > > >
> > > >     admin.enableTable(TABLENAME);
> > > >     // Assert that every regionserver has some regions on it.
> > > >     MiniHBaseCluster cluster = TEST_UTIL.getMiniHBaseCluster();
> > > >     for (int i = 0; i < cluster.getRegionServerThreads().size();
> i++) {
> > > >       HRegionServer hrs = cluster.getRegionServer(i);
> > > >       Assert.assertFalse(hrs.getOnlineRegions().isEmpty());
> > > >     }
> > > >
> > > > When table is enabled, it was possible that some region server didn't
> > > carry
> > > > any region.
> > > > It seems the above assertion can be removed.
> > > >
> > > > Cheers
> > > >
> > > > On Wed, Aug 29, 2012 at 11:14 PM, Stack <[EMAIL PROTECTED]> wrote:
> > > >
> > > > > On Wed, Aug 29, 2012 at 11:03 AM, Ted Yu <[EMAIL PROTECTED]>
> > wrote:
> > > > > > According to:
> > > > > >
> > > > >
> > > >
> > >
> >
> https://builds.apache.org/view/G-L/view/HBase/job/HBase-0.92/519/testReport/junit/org.apache.hadoop.hbase.regionserver/TestSplitTransactionOnCluster/testSplitBeforeSettingSplittingInZK/
> > > > > >
> > > > > > The test failure was due to:
> > > > > >
> > > > > > java.lang.RuntimeException: java.io.FileNotFoundException:
> > > > > >
> > > > >
> > > >
> > >
> >
> /home/hudson/hudson-slave/workspace/HBase-0.92/trunk/target/classes/hbase-default.xml
> > > > > > (Too many open files)
> > > > > >
> > > > > > I think currently only HBASE-6649 (TestReplication.queueFailover)
> > is
> > > > open
> > > > > >
> > > > > > TestReplication actually hangs in trunk. So it should be fixed
> > across
> > > > > 0.92,
> > > > > > 0.94 and trunk.
> > > > > >
> > > > >
> > > > > I haven't seen 'too many open files' in recent runs.
> > > > >
> > > > > I saw TestDrainingServer fail here
> > > > > https://builds.apache.org/view/G-L/view/HBase/job/HBase-0.92/543/
> > > > > weirdly in setUpBeforeClass
> > > > >
> > > > > TestSplitLogManager here:
> > > > > https://builds.apache.org/view/G-L/view/HBase/job/HBase-0.92/542/
> > > > >
> > > > > It seems like TestMetaReaderEditor failed in 540.....
> > > > >
> > > > > Lets try fix a few of the above such that 0.92 passes more often
> that
> > > > > it fails (Its better already w/ fixes that have gone in lately in
> > that
> > > > > it usually passes when I do a build locally).
> > > > >
> > > > > St.Ack
> > > > >
> > > >
> > >
> >
>