[ 
https://issues.apache.org/jira/browse/BEAM-4770?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

Lukasz Gajowy updated BEAM-4770:
--------------------------------
    Description: 
+What was the setup?+

I created Pull Request (let's call it "A") of a branch with some changes in 
job__00_seed.groovy. Those changes involved adding a new glob pattern to detect 
files not only with job_*.groovy naming pattern but also with /**/*Job.groovy.

I created another PR (let's call it "B") which was "empty" - no changes added 
to the codebase. I used such PR to run job_seed_standalone (or job_00_seed) 
from master branch to revert changes introduced by A.

So:

 - to introduce change I ran: "Run seed job" in pr A's comment

 - to revert changes I ran: "Run seed job" / "Run seed job standalone" in pr 
B's comment. 

 

+What did I expect?+

I expected every change introduced in the seed job by A reverted by B's 
standalone seed job. I even expected that B's seed job will revert the changes.

+What actually happened?+

After running standalone seed job from "B" some changes were not reverted. 
Jenkins was still expecting files with *Job. glob pattern when I was running 
the seed job afterwards. It couldn't find them on master so it resulted in the 
following error: 
{code:java}
Resetting working tree
 > git reset --hard # timeout=10
 > git clean -fdx # timeout=10
ERROR: no Job DSL script(s) found at .test-infra/jenkins/**/*Job.groovy
Not sending mail to unregistered user ccla...@bluewin.ch
Sending e-mails to: d...@beam.apache.org pabl...@google.com
Setting status of 4715449665ae235f96c7a67ce5f80b774d5474e5 to FAILURE with url 
https://builds.apache.org/job/beam_SeedJob/2191/ and message: 'FAILURE
 '
Using context: Jenkins: Seed Job
Finished: FAILURE
{code}
 

+Enclosed Links:+


 # Original Pull Request A (no. 5914)
 # Original Pull Request B (no. 5915)

  was:
+What was the setup?+

I created Pull Request (let's call it "A") of a branch with some changes in 
job__00_seed.groovy. Those changes involved adding a new glob pattern to detect 
files not only with job_*.groovy naming pattern but also with /**/*Job.groovy.

I created another PR (let's call it "B") which was "empty" - no changes added 
to the codebase. I used such PR to run job_seed_standalone (or job_00_seed) 
from master branch to revert changes introduced by A.

So:

 - to introduce change I ran: "Run seed job" in pr A's comment

 - to revert changes I ran: "Run seed job" / "Run seed job standalone" in pr 
B's comment. 

 

+What did I expect?+

I expected every change introduced in the seed job by A reverted by B's 
standalone seed job. I even expected that B's seed job will revert the changes.

+What actually happened?+

After running standalone seed job from "B" some changes were not reverted. 
Jenkins was still expecting files with *Job. glob pattern when I was running 
the seed job afterwards. It couldn't find them on master so it resulted in the 
following error: 


{code:java}
Resetting working tree
 > git reset --hard # timeout=10
 > git clean -fdx # timeout=10
ERROR: no Job DSL script(s) found at .test-infra/jenkins/**/*Job.groovy
Not sending mail to unregistered user ccla...@bluewin.ch
Sending e-mails to: d...@beam.apache.org pabl...@google.com
Setting status of 4715449665ae235f96c7a67ce5f80b774d5474e5 to FAILURE with url 
https://builds.apache.org/job/beam_SeedJob/2191/ and message: 'FAILURE
 '
Using context: Jenkins: Seed Job
Finished: FAILURE
{code}

+Enclosed Links:
+
 # Original Pull Request A (no. 5914)
 # Original Pull Request B (no. 5915)


> Standalone seed job didn't revert Jenkins seed job changes introduced by a PR
> -----------------------------------------------------------------------------
>
>                 Key: BEAM-4770
>                 URL: https://issues.apache.org/jira/browse/BEAM-4770
>             Project: Beam
>          Issue Type: Bug
>          Components: testing
>            Reporter: Lukasz Gajowy
>            Assignee: Jason Kuster
>            Priority: Major
>
> +What was the setup?+
> I created Pull Request (let's call it "A") of a branch with some changes in 
> job__00_seed.groovy. Those changes involved adding a new glob pattern to 
> detect files not only with job_*.groovy naming pattern but also with 
> /**/*Job.groovy.
> I created another PR (let's call it "B") which was "empty" - no changes added 
> to the codebase. I used such PR to run job_seed_standalone (or job_00_seed) 
> from master branch to revert changes introduced by A.
> So:
>  - to introduce change I ran: "Run seed job" in pr A's comment
>  - to revert changes I ran: "Run seed job" / "Run seed job standalone" in pr 
> B's comment. 
>  
> +What did I expect?+
> I expected every change introduced in the seed job by A reverted by B's 
> standalone seed job. I even expected that B's seed job will revert the 
> changes.
> +What actually happened?+
> After running standalone seed job from "B" some changes were not reverted. 
> Jenkins was still expecting files with *Job. glob pattern when I was running 
> the seed job afterwards. It couldn't find them on master so it resulted in 
> the following error: 
> {code:java}
> Resetting working tree
>  > git reset --hard # timeout=10
>  > git clean -fdx # timeout=10
> ERROR: no Job DSL script(s) found at .test-infra/jenkins/**/*Job.groovy
> Not sending mail to unregistered user ccla...@bluewin.ch
> Sending e-mails to: d...@beam.apache.org pabl...@google.com
> Setting status of 4715449665ae235f96c7a67ce5f80b774d5474e5 to FAILURE with 
> url https://builds.apache.org/job/beam_SeedJob/2191/ and message: 'FAILURE
>  '
> Using context: Jenkins: Seed Job
> Finished: FAILURE
> {code}
>  
> +Enclosed Links:+
>  # Original Pull Request A (no. 5914)
>  # Original Pull Request B (no. 5915)



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to