Re: Very red CI -> Hold merges, please

I merged it in. On Thu, Feb 7, 2019 at 11:28 AM Kirk Lund wrote: > I think we should go ahead and merge in > https://github.com/apache/geode/pull/3172 since it resolves the > GfshConsoleModeUnitTest UnitTest failures. > > On Thu, Feb 7, 2019 at 9:57 AM Nabarun Nag wrote: > >> FYI, I have just

Re: Very red CI -> Hold merges, please

Sorry, go ahead and revert the commit and reopen the PR. On Thu, Feb 7, 2019 at 11:36 AM Jens Deppe wrote: > I was still working on a fix... > > On Thu, Feb 7, 2019 at 11:31 AM Kirk Lund wrote: > > > I merged it in. > > > > On Thu, Feb 7, 2019 at 11:28 AM Kirk Lund wrote: > > > > > I think we

Re: Very red CI -> Hold merges, please

FYI, I have just merged a ci timeout fix to increase the timeout for geode-benchmarks to 4h. This does not influence any geode modules. Regards Naba On Thu, Feb 7, 2019 at 9:32 AM Alexander Murmann wrote: > Hi folks, > > Our CI is very red since ~24 hours > < >

Very red CI -> Hold merges, please

Hi folks, Our CI is very red since ~24 hours . It looks like a substantial new issue was introduced. Can we hold off on merging new changes to the develop branch till this issue is

Re: Very red CI -> Hold merges, please

I think we should go ahead and merge in https://github.com/apache/geode/pull/3172 since it resolves the GfshConsoleModeUnitTest UnitTest failures. On Thu, Feb 7, 2019 at 9:57 AM Nabarun Nag wrote: > FYI, I have just merged a ci timeout fix to increase the timeout for > geode-benchmarks to 4h.

Re: Very red CI -> Hold merges, please

I was still working on a fix... On Thu, Feb 7, 2019 at 11:31 AM Kirk Lund wrote: > I merged it in. > > On Thu, Feb 7, 2019 at 11:28 AM Kirk Lund wrote: > > > I think we should go ahead and merge in > > https://github.com/apache/geode/pull/3172 since it resolves the > > GfshConsoleModeUnitTest

Re: Very red CI -> Hold merges, please

I'm looking into the locator test failures.  I recently changed those tests to make them not depend on the old JUnit4DistributedTestCase and they lost some settings along the way.  There's also a real problem with Geode that this test is exposing because of that change, so I'd like to NOT fix

Re: Very red CI -> Hold merges, please

Hmm, and that was another false search hit in Jira! Searching for WANRollingUpgradeNewSenderProcessOldEvent in Jira brings up GEODE-3967 which apparently does NOT involve that test. So, maybe we found another flaky test. Jira search seems to not work very well. On Thu, Feb 7, 2019 at 2:24 PM

Re: Very red CI -> Hold merges, please

No worries. I think I have a better fix now. At least the builds are moving again. On Thu, Feb 7, 2019 at 12:11 PM Kirk Lund wrote: > Sorry, go ahead and revert the commit and reopen the PR. > > On Thu, Feb 7, 2019 at 11:36 AM Jens Deppe wrote: > > > I was still working on a fix... > > > > On

Re: Very red CI -> Hold merges, please

The UpgradeTest failures on your latest commit for this PR are WANRollingUpgradeNewSenderProcessOldEvent which seems to be a reoccurrence of [GEODE-3967](https://issues.apache.org/jira/browse/GEODE-3967). I recommend having Gester take a look at that these failures. He marked

Re: Very red CI -> Hold merges, please

Bruce, would it make sense to for now revert the suspect change to the test? At that point we should be back to full green and we all can without a doubt go back to our usual flow of merging to develop. Thoughts? On Thu, Feb 7, 2019 at 2:37 PM Kirk Lund wrote: > Hmm, and that was another false

Re: Very red CI -> Hold merges, please

Awesome! Things look a lot better now. It seems like the only consistent issue now is in DistributedTestOpenJDK11 where the LocatorDUnitTest `testMultipleLocatorsRestartingAtSameTimeWithMissingServers` has been failing since 2 days or so. Is anyone looking at that? On Thu, Feb 7, 2019 at 12:37 PM

Adding integrationTest src set to geode-dunit

The usual geode src sets like integrationTest don't already exist in some modules such as geode-dunit. I'm trying to write a new IntegrationTest but simply creating the directories and placing a new .java file in it doesn't seem to work.

Re: Adding integrationTest src set to geode-dunit

Can I work with you on this tomorrow? On Thu, Feb 7, 2019, 15:09 Kirk Lund wrote: > The usual geode src sets like integrationTest don't already exist in some > modules such as geode-dunit. > > I'm trying to write a new IntegrationTest but simply creating the > directories and placing a new

Re: Very red CI -> Hold merges, please

Pipeline is back to green now. Thank you to everyone who stepped up to get things back on track. If you had PR checks fail this week, please re-trigger them (by making an empty commit). > On Feb 7, 2019, at 4:20 PM, Alexander Murmann wrote: > > Bruce, would it make sense to for now revert

Re: Very red CI -> Hold merges, please

WANRollingUpgradeNewSenderProcessOldEvent is not related with GEODE-3967. I wonder why search guided us to GEODE-3967. Regards Gester On Thu, Feb 7, 2019 at 8:34 PM Owen Nichols wrote: > Pipeline is back to green now. Thank you to everyone who stepped up to > get things back on track. > > If

Re: GfshConsoleModeUnitTest is failing in UnitTest

The problem is caused by some change in test ordering (unit tests are running in parallel) which is resulting in environmental pollution. Probably Government shutdown related... :P On Wed, Feb 6, 2019 at 7:31 PM Jens Deppe wrote: > I didn't touch anything in this area, but nevertheless I

Re: GfshConsoleModeUnitTest is failing in UnitTest

Oh, feel free to review https://github.com/apache/geode/pull/3172. Better suggestions on how to 'fix' the problem are most welcome. On Thu, Feb 7, 2019 at 6:49 AM Jens Deppe wrote: > The problem is caused by some change in test ordering (unit tests are > running in parallel) which is resulting