Status: Resolved (View Workflow)
Blue Ocean 1.5
Steps to reproduce using two jobs:
- First job: triggers the second job (wait = true)
- Second job: record unit tests
- Open Blue Ocean from the first job, click Triggered Builds to access the second job
- The Tests tab will say no results are recorded
- Exit Blue Ocean, and access it again from the second job
- The Tests tab will show results
I noticed this happening both while the job was running and complete.
- is related to
JENKINS-51110 no test results visible in blueocean when following link from classic -> BO
JENKINS-55497 BlueOcean Pipeline View for the Latest Run shows No Changes
stuartrowe I'm confused by your comment. Sounds like you suspect its not fixed, but reinholdfuereder says it is actually fixed for them?
I agree, the null looks suspicious, but the PR it was merged in with (https://github.com/jenkinsci/blueocean-plugin/commit/1e05a5cf487cdb85112cbbb61d6ebf1f9cb889b4) says it was done for performance issues.
I think a longer term better fix is making the tests page pull the data it needs if its not available, but i'm way less familiar with the actual frontend code here.
I'm just saying that if it is fixed, it's not due to my PR. I looped you in in case there were any other recent BO changes that came to mind. I don't have jobs that produce xUnit style output so I haven't noticed this issue and can't confirm a fix.
Okay cool, I agree, i don't think its related to your change, but its a complicated system so who knows.
I'm leaving this as resolved!
stuartrowe and halkeye: Sorry for the confusion. I am really extremely pretty sure that the problem I described in https://issues.jenkins-ci.org/browse/JENKINS-51111?focusedCommentId=343321&page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel#comment-343321 (with rather straightforward reproducer steps) works now as expected (for "changes" AND for "tests") – and I think that was really the same problem as this issue's original description.
However, I could also imagine that stuartrowe is perfectly right about that he only fixed the "changes" now in the course of
JENKINS-55497, and "tests" was already fixed some time ago...
I don't believe my change would have fixed this issue. It sounds like a similar bug though (retrieving and caching the run summary for the last run of a project leads to false cache hit when retrieving the full details for the same run). This code snippet looks suspicious:
halkeye could you take a look / comment on this?