[GRADLE-2035] Error running tests in milestone 6 Created: 06/Jan/12 Updated: 04/Jan/13 Resolved: 23/Jan/12 |
|
Status: | Resolved |
Project: | Gradle |
Affects Version/s: | 1.0-milestone-6 |
Fix Version/s: | 1.0-milestone-8 |
Type: | Bug | ||
Reporter: | Gradle Forums | Assignee: | Szczepan Faber |
Resolution: | Fixed | Votes: | 5 |
Attachments: | TEST-files.zip |
Description |
I recently switched to milestone 6 and started getting this error when running tests in one of my modules: Unable to notify listener with output event: org.gradle.api.internal.tasks.testing.DefaultTestOutputEvent@68f2a728. I couldn't find test with id: 15.108. This might be a bug, please report it. I'm not even sure which test it's failing on because that's the only information I get, otherwise I'd post the test example. This was not a problem in milestone 3 with the exact same tests and dependencies. I'm using Java JDK 1.6.0_29-b11 on a Linux Ubuntu machine. Any help would be greatly appreciated as it's killing out CI machine, which has been updated to milestone 6. Thanks! |
Comments |
Comment by Gradle Forums [ 06/Jan/12 ] |
Most likely this isn't a problem with a particular test but with Gradle's testing infrastructure. It would be very valuable if you could provide a build that allows to reproduce the problem. What are your test settings? Do you run tests in parallel? |
Comment by Gradle Forums [ 06/Jan/12 ] |
I'll try and put together a test that causes this issue. But where do I find the test id's that Gradle assigns to each test? When it says: I couldn't find test with id: 1.108 How do I find out which test it assigned as 1.108? |
Comment by Chris Beams [ 09/Jan/12 ] |
Just a friendly reminder to see about scheduling this for 1.0-milestone-8, so it doesn't slip through the cracks. Thanks! |
Comment by Adam Murdoch [ 09/Jan/12 ] |
@chris, we're working on this. It's more or less fixed in master. |
Comment by Chris Beams [ 10/Jan/12 ] |
Right; Szczepan pointed me at a snapshot and I'm trying it out now. |
Comment by Chris Beams [ 12/Jan/12 ] |
I'm now running successfully against gradle-1.0-milestone-8-20120112000036+0100. I've got a build plan in a loop running every fifteen minutes to see if we hit the intermittent "couldn't find test with id" error. If it crops back up, I'll comment here. In the meantime, we are still getting warnings like these – it seems related: Failed to parse test result file "TEST-org.springframework.integration.amqp.config.OutboundGatewayIntegrationTests.xml" Failed to parse test result file "TEST-org.springframework.integration.aggregator.AggregatorTests.xml" Failed to parse test result file "TEST-org.springframework.integration.aggregator.ConcurrentAggregatorTests.xml" Failed to parse test result file "TEST-org.springframework.integration.aggregator.scenarios.AggregationResendTests.xml" Failed to parse test result file "TEST-org.springframework.integration.core.MessageIdGenerationTests.xml" Failed to parse test result file "TEST-org.springframework.integration.dispatcher.AggregateMessageDeliveryExceptionTests.xml" Failed to parse test result file "TEST-org.springframework.integration.gateway.MessagingGatewayTests.xml" Failed to parse test result file "TEST-org.springframework.integration.history.MessageHistoryIntegrationTests.xml" Failed to parse test result file "TEST-org.springframework.integration.store.MessageGroupQueueTests.xml" Failed to parse test result file "TEST-org.springframework.integration.feed.config.FeedInboundChannelAdapterParserTests.xml" ... See https://build.springsource.org/browse/INT-BUILD-JOB2-11 Any idea why? And how to get rid of them? Thanks for the fixes so far! |
Comment by Chris Beams [ 12/Jan/12 ] |
As mentioned above, I put https://build.springsource.org/browse/INT-BUILD/ in a loop against the latest m8 snapshot, and after twenty or so builds, have seen no failures. So it looks like this issue is indeed taken care of. Thanks! Any comment on the "Failed to parse test result" entries would still be appreciated, as they are still occurring throughout all those recent runs. |
Comment by Szczepan Faber [ 12/Jan/12 ] |
Looking at it... |
Comment by Szczepan Faber [ 12/Jan/12 ] |
It appears this exception comes from Bamboo. It is a problem because you don't have reports available for those tests in Bamboo :/ How can I get hold of those 'invalid' xmls? |
Comment by Chris Beams [ 12/Jan/12 ] |
Attached all TEST-* files from the most recent run of INT-BUILD |
Comment by Szczepan Faber [ 13/Jan/12 ] |
All files look good to me, they contain all data, they are valid xml, etc. I don't know why bamboo refuses to parse them. Perhaps something else has changed in the environment that might introduce this behavior? Are you sure this error wasn't there before? Might be worth looking closer into bamboo logs or ask the atlassian dudes why a valid report cannot be parsed (or suggest them to add some information to the warning why the file cannot be parsed). It's hard to be 100% sure these days but I'm reasonably confident my change couldn't break the xml generation E.g. implementation wise the xml generator is one TestResultProcessor and I was changing a different TestResultProcessor. |
Comment by Chris Beams [ 13/Jan/12 ] |
Actually, looking back in the build logs, these warnings have been happening for quite some time. Here's one from November: https://build.springsource.org/browse/INT-MASTER-JOB1-198 So it appears that this really is a separate issue from the "couldn't find test with id" problem that was addressed here. The SI guys are all watching this issue, so I'll pass your advice above along to them:
This is something that might be worth taking up with our Atlassian support contract, but in the meantime, it's clearly benign with regard to the SI build, and seems only to happen under CI. Probably not so critical, then. Thanks, Szczepan! Please resolve at will; we're looking good now. |
Comment by Szczepan Faber [ 13/Jan/12 ] |
Cool |