I opened up ATLAS-2096 on the Dimension already exists error which I continue to hit today after pulling the latest master, and I also see intermittently in the Jenkins build.
On 2017-08-25 19:10, Sarath Subramanian <[email protected]> wrote: > Nigel, > I agree we all need to spend some time and clean up these tests. Feel free > to open JIRA's for the failing tests. > > I see the integration tests consistently failing in hive-bridge module due > to Lock timeout exception - I suspect this is because of some stale open > transaction lingering around which also modified/read the same property, > since Berkley DB is ACID compliant subsequent transactions times out since > the previous transaction is not closed properly. We don't see this in HBase > since its not an ACID compliant. > > > Thanks, > Sarath Subramanian > > On Fri, Aug 25, 2017 at 7:10 AM, Apoorv Naik <[email protected]> wrote: > > > I think Sarath already made changes for this one don't know why this > > failed again. Perhaps we need to revisit the integration tests after all. > > > > One jira per test failure should be a great way to split work amongst the > > community. > > > > Sent from Nine<http://www.9folders.com/> > > ________________________________ > > From: Nigel Jones <[email protected]> > > Sent: Aug 25, 2017 5:06 AM > > To: [email protected] > > Subject: Re: Emtity Resource v1 test failures > > > > > > On 2017-08-25 12:28, "Nigel Jones"<[email protected]> wrote: > > > > > I rebased this morning, but am still getting a failure - albeit a > > different one - this looks like a timeout issue whilst the previous were > > seemingly more functional. Is there a JIRA you are tracking these under? > > Should I create one per failure? > > > > > > I'll retry > > > > > > Tests run: 133, Failures: 1, Errors: 0, Skipped: 0, Time elapsed: > > 109.969 sec <<< FAILURE! - in TestSuite > > > testLineageIsMaintained(org.apache.atlas.examples.QuickStartIT) Time > > elapsed: 1.537 sec <<< FAILURE! > > > > Another build, another random error - timing could be a factor (certainly > > in the earlier test). This however is one of the build failures I had > > yesterday. It's also been seen in the Jenkins build > > > > FAILURE! - in TestSuite > > runQuickStart(org.apache.atlas.examples.QuickStartV2IT) Time elapsed: > > 0.344 sec <<< FAILURE! > > org.apache.atlas.AtlasServiceException: Metadata service API > > org.apache.atlas.AtlasBaseClient$APIInfo@1a990d3 failed with status 409 > > (Conflict) Response Body > > ({"errorCode":"ATLAS-409-00-001","errorMessage":"Given > > type Dimension already exists"}) > > at org.apache.atlas.examples.QuickStartV2IT.runQuickStart( > > QuickStartV2IT.java:49) > > > > > > Results : > > > > Failed tests: > > QuickStartV2IT.runQuickStart:49 û AtlasService Metadata service API > > org.apache... > > > > Tests run: 134, Failures: 1, Errors: 0, Skipped: 5 > > > > > > I feel we do need to between us all try and get these tests running > > cleanly, as it's otherwise very tempting to run skip-tests and risk > > suggesting a change/patch that causes a failure. (I was trying to get a > > solid baseline before trying to start on a new API) > > > > We could even consider a short term change to the build to not stop if any > > known, problematic tests fail - still run, but continue. may help? > > > > >
