HBase, mail # dev - Hudson fails most of the time

Re: Hudson fails most of the time

Andrew Purtell 2012-08-16, 18:42

Great to see. I've been fixing tests too as have the available bandwidth.
We should get Hudson back to green on all projects.

There are some new failures showing up on recent branch-2 + 0.94.

We're testing an internal fix for TestImportExport. I know Jon committed
then reverted a fix for that, seems copying over the yarn.* configuration
items is sufficient and won't break Hadoop 1. Our new guy Scott Forman has
been working on this.

Aside from that, also TestReplication, TestMiniClusterLoadParallel (test
timeouts), TestMetaMigrationRemovingHTD, and
TestLogRolling.testLogRollOnPipelineRestart (which may need HDFS side
help).

I have our new guys working on these to learn the ropes. We're going to cut
at the end of this quarter and stabilize in Q4 to release at the end of the
year of our new platform. Until then we're following head of Hadoop
branch-2 and head of HBase 0.94 branch. First item to tackle is getting
green builds. We'll contribute as we can.

- Andy

On Thu, Aug 16, 2012 at 11:34 AM, lars hofhansl <[EMAIL PROTECTED]> wrote:

> We're starting an initiative on this here at Salesforce. We'll be happy to
> work together on this.
>
>
>
> ________________________________
> From: Jimmy Xiang <[EMAIL PROTECTED]>
> To: [EMAIL PROTECTED]
> Sent: Thursday, August 16, 2012 11:14 AM
> Subject: Hudson fails most of the time
>
> Should we spend some time to fix those flaky tests? Before they are
> not flaky any more, should we disable them for now?
>
> Thanks,
> Jimmy
>

--
Best regards,

- Andy

Problems worthy of attack prove their worth by hitting back. - Piet Hein
(via Tom White)

Re: Hudson fails most of the time

Jimmy Xiang 2012-08-16, 18:52

That's really great!

Probably we can file a Jira for each failed test case. So that
we don't duplicate the efforts.

Thanks,
Jimmy

On Thu, Aug 16, 2012 at 11:42 AM, Andrew Purtell <[EMAIL PROTECTED]> wrote:
> Great to see. I've been fixing tests too as have the available bandwidth.
> We should get Hudson back to green on all projects.
>
> There are some new failures showing up on recent branch-2 + 0.94.
>
> We're testing an internal fix for TestImportExport. I know Jon committed
> then reverted a fix for that, seems copying over the yarn.* configuration
> items is sufficient and won't break Hadoop 1. Our new guy Scott Forman has
> been working on this.
>
> Aside from that, also TestReplication, TestMiniClusterLoadParallel (test
> timeouts), TestMetaMigrationRemovingHTD, and
> TestLogRolling.testLogRollOnPipelineRestart (which may need HDFS side
> help).
>
> I have our new guys working on these to learn the ropes. We're going to cut
> at the end of this quarter and stabilize in Q4 to release at the end of the
> year of our new platform. Until then we're following head of Hadoop
> branch-2 and head of HBase 0.94 branch. First item to tackle is getting
> green builds. We'll contribute as we can.
>
> - Andy
>
> On Thu, Aug 16, 2012 at 11:34 AM, lars hofhansl <[EMAIL PROTECTED]> wrote:
>
>> We're starting an initiative on this here at Salesforce. We'll be happy to
>> work together on this.
>>
>>
>>
>> ________________________________
>> From: Jimmy Xiang <[EMAIL PROTECTED]>
>> To: [EMAIL PROTECTED]
>> Sent: Thursday, August 16, 2012 11:14 AM
>> Subject: Hudson fails most of the time
>>
>> Should we spend some time to fix those flaky tests? Before they are
>> not flaky any more, should we disable them for now?
>>
>> Thanks,
>> Jimmy
>>
>
>
>
> --
> Best regards,
>
> - Andy
>
> Problems worthy of attack prove their worth by hitting back. - Piet Hein
> (via Tom White)

Re: Hudson fails most of the time

Dave Wang 2012-08-16, 19:01

Sounds like a good idea. Our team has been fixing some of these flaky
tests, and I'm glad to hear other folks have been looking at this as well.

Once we get everything back to green, we should focus on keeping them
green, as otherwise the tests tend to get ignored.

- Dave

On Thu, Aug 16, 2012 at 11:52 AM, Jimmy Xiang <[EMAIL PROTECTED]> wrote:

> That's really great!
>
> Probably we can file a Jira for each failed test case. So that
> we don't duplicate the efforts.
>
> Thanks,
> Jimmy
>
> On Thu, Aug 16, 2012 at 11:42 AM, Andrew Purtell <[EMAIL PROTECTED]>
> wrote:
> > Great to see. I've been fixing tests too as have the available bandwidth.
> > We should get Hudson back to green on all projects.
> >
> > There are some new failures showing up on recent branch-2 + 0.94.
> >
> > We're testing an internal fix for TestImportExport. I know Jon committed
> > then reverted a fix for that, seems copying over the yarn.* configuration
> > items is sufficient and won't break Hadoop 1. Our new guy Scott Forman
> has
> > been working on this.
> >
> > Aside from that, also TestReplication, TestMiniClusterLoadParallel (test
> > timeouts), TestMetaMigrationRemovingHTD, and
> > TestLogRolling.testLogRollOnPipelineRestart (which may need HDFS side
> > help).
> >
> > I have our new guys working on these to learn the ropes. We're going to
> cut
> > at the end of this quarter and stabilize in Q4 to release at the end of
> the
> > year of our new platform. Until then we're following head of Hadoop
> > branch-2 and head of HBase 0.94 branch. First item to tackle is getting
> > green builds. We'll contribute as we can.
> >
> > - Andy
> >
> > On Thu, Aug 16, 2012 at 11:34 AM, lars hofhansl <[EMAIL PROTECTED]>
> wrote:
> >
> >> We're starting an initiative on this here at Salesforce. We'll be happy
> to
> >> work together on this.
> >>
> >>
> >>
> >> ________________________________
> >> From: Jimmy Xiang <[EMAIL PROTECTED]>
> >> To: [EMAIL PROTECTED]
> >> Sent: Thursday, August 16, 2012 11:14 AM
> >> Subject: Hudson fails most of the time
> >>
> >> Should we spend some time to fix those flaky tests? Before they are
> >> not flaky any more, should we disable them for now?
> >>
> >> Thanks,
> >> Jimmy
> >>
> >
> >
> >
> > --
> > Best regards,
> >
> > - Andy
> >
> > Problems worthy of attack prove their worth by hitting back. - Piet Hein
> > (via Tom White)
>