You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@sentry.apache.org by Sravya Tirukkovalur <sr...@cloudera.com> on 2014/08/27 02:08:38 UTC
Lets keep precommit build in a reliable state
Hi Sentry contributors,
We have had our pre commit build pretty unstable for a while. After some
test fixes and improved bigtop slaves health, it is back to being healthy
now.
http://bigtop01.cloudera.org:8080/job/PreCommit-SENTRY-Build/ (All red ones
are actual patch related test failures).
Lets take this opportunity to make "all tests passing" on precommit build a
requirement for every patch. Of course, we will file jiras for un related
flaky tests and prioritize them to unblock the commits if need be. Any
thoughts?
Thanks!
--
Sravya Tirukkovalur
Re: Lets keep precommit build in a reliable state
Posted by Sravya Tirukkovalur <sr...@cloudera.com>.
Thanks for the responses folks! At this point the only known reason for
false failures apart from machine issues seem to be
https://issues.apache.org/jira/browse/SENTRY-208. (8 out of 10 last false
failures, other 2 I think are machine issues). Once we fix this issue, we
should have more stable runs (patch is available). But I agree that lets
try to keep the pre commit build green as much as possible, but still allow
non pre commit validation by patch provider and committer when necessary.
Thanks!
On Tue, Aug 26, 2014 at 5:24 PM, Prasad Mujumdar <pr...@cloudera.com>
wrote:
> If you re-attach the same patch, it should kick off a new build.
>
> thanks
> Prasad
>
>
>
> On Tue, Aug 26, 2014 at 5:22 PM, Gregory Chanan <gc...@cloudera.com>
> wrote:
>
> > Is there an easy way to kick off another pre commit build in case a flaky
> > issue occurs?
> >
> > Greg
> >
> >
> > On Tue, Aug 26, 2014 at 5:17 PM, Prasad Mujumdar <pr...@cloudera.com>
> > wrote:
> >
> > > Thanks for taking initiative to drive this!
> > > Sounds good to me, though it's always possible to see unrelated
> failures
> > > due to machine issues, possible flaky tests. Perhaps we should allow a
> > > followup re-validation of failed tests by the fixer.
> > >
> > > thanks
> > > Prasad
> > >
> > >
> > >
> > > On Tue, Aug 26, 2014 at 5:08 PM, Sravya Tirukkovalur <
> > sravya@cloudera.com>
> > > wrote:
> > >
> > > > Hi Sentry contributors,
> > > >
> > > > We have had our pre commit build pretty unstable for a while. After
> > some
> > > > test fixes and improved bigtop slaves health, it is back to being
> > healthy
> > > > now.
> > > > http://bigtop01.cloudera.org:8080/job/PreCommit-SENTRY-Build/ (All
> red
> > > > ones
> > > > are actual patch related test failures).
> > > >
> > > > Lets take this opportunity to make "all tests passing" on precommit
> > > build a
> > > > requirement for every patch. Of course, we will file jiras for un
> > related
> > > > flaky tests and prioritize them to unblock the commits if need be.
> Any
> > > > thoughts?
> > > >
> > > > Thanks!
> > > > --
> > > > Sravya Tirukkovalur
> > > >
> > >
> >
>
--
Sravya Tirukkovalur
Re: Lets keep precommit build in a reliable state
Posted by Prasad Mujumdar <pr...@cloudera.com>.
If you re-attach the same patch, it should kick off a new build.
thanks
Prasad
On Tue, Aug 26, 2014 at 5:22 PM, Gregory Chanan <gc...@cloudera.com>
wrote:
> Is there an easy way to kick off another pre commit build in case a flaky
> issue occurs?
>
> Greg
>
>
> On Tue, Aug 26, 2014 at 5:17 PM, Prasad Mujumdar <pr...@cloudera.com>
> wrote:
>
> > Thanks for taking initiative to drive this!
> > Sounds good to me, though it's always possible to see unrelated failures
> > due to machine issues, possible flaky tests. Perhaps we should allow a
> > followup re-validation of failed tests by the fixer.
> >
> > thanks
> > Prasad
> >
> >
> >
> > On Tue, Aug 26, 2014 at 5:08 PM, Sravya Tirukkovalur <
> sravya@cloudera.com>
> > wrote:
> >
> > > Hi Sentry contributors,
> > >
> > > We have had our pre commit build pretty unstable for a while. After
> some
> > > test fixes and improved bigtop slaves health, it is back to being
> healthy
> > > now.
> > > http://bigtop01.cloudera.org:8080/job/PreCommit-SENTRY-Build/ (All red
> > > ones
> > > are actual patch related test failures).
> > >
> > > Lets take this opportunity to make "all tests passing" on precommit
> > build a
> > > requirement for every patch. Of course, we will file jiras for un
> related
> > > flaky tests and prioritize them to unblock the commits if need be. Any
> > > thoughts?
> > >
> > > Thanks!
> > > --
> > > Sravya Tirukkovalur
> > >
> >
>
Re: Lets keep precommit build in a reliable state
Posted by Gregory Chanan <gc...@cloudera.com>.
Is there an easy way to kick off another pre commit build in case a flaky
issue occurs?
Greg
On Tue, Aug 26, 2014 at 5:17 PM, Prasad Mujumdar <pr...@cloudera.com>
wrote:
> Thanks for taking initiative to drive this!
> Sounds good to me, though it's always possible to see unrelated failures
> due to machine issues, possible flaky tests. Perhaps we should allow a
> followup re-validation of failed tests by the fixer.
>
> thanks
> Prasad
>
>
>
> On Tue, Aug 26, 2014 at 5:08 PM, Sravya Tirukkovalur <sr...@cloudera.com>
> wrote:
>
> > Hi Sentry contributors,
> >
> > We have had our pre commit build pretty unstable for a while. After some
> > test fixes and improved bigtop slaves health, it is back to being healthy
> > now.
> > http://bigtop01.cloudera.org:8080/job/PreCommit-SENTRY-Build/ (All red
> > ones
> > are actual patch related test failures).
> >
> > Lets take this opportunity to make "all tests passing" on precommit
> build a
> > requirement for every patch. Of course, we will file jiras for un related
> > flaky tests and prioritize them to unblock the commits if need be. Any
> > thoughts?
> >
> > Thanks!
> > --
> > Sravya Tirukkovalur
> >
>
Re: Lets keep precommit build in a reliable state
Posted by Prasad Mujumdar <pr...@cloudera.com>.
Thanks for taking initiative to drive this!
Sounds good to me, though it's always possible to see unrelated failures
due to machine issues, possible flaky tests. Perhaps we should allow a
followup re-validation of failed tests by the fixer.
thanks
Prasad
On Tue, Aug 26, 2014 at 5:08 PM, Sravya Tirukkovalur <sr...@cloudera.com>
wrote:
> Hi Sentry contributors,
>
> We have had our pre commit build pretty unstable for a while. After some
> test fixes and improved bigtop slaves health, it is back to being healthy
> now.
> http://bigtop01.cloudera.org:8080/job/PreCommit-SENTRY-Build/ (All red
> ones
> are actual patch related test failures).
>
> Lets take this opportunity to make "all tests passing" on precommit build a
> requirement for every patch. Of course, we will file jiras for un related
> flaky tests and prioritize them to unblock the commits if need be. Any
> thoughts?
>
> Thanks!
> --
> Sravya Tirukkovalur
>