Skip to content

Conversation

@ndokos
Copy link
Member

@ndokos ndokos commented Nov 24, 2015

Some versions of fio produce an extra disk-stats array.
This change makes the postprocess script ignore the extra array
if present.

Some versions of fio produce an extra disk-stats array.
This change makes the postprocess script ignore the extra array
if present.
@ndokos ndokos force-pushed the wip-fio-postprocess branch from 3ca21f4 to 4eba91b Compare November 24, 2015 20:29
ndokos added a commit that referenced this pull request Nov 24, 2015
Ignore extra disk-stats array if present.
@ndokos ndokos merged commit 5074425 into distributed-system-analysis:master Nov 24, 2015
@ndokos ndokos deleted the wip-fio-postprocess branch December 2, 2015 18:46
portante pushed a commit that referenced this pull request Oct 16, 2021
Starting with the 30 September 2021 build ([#110](http://jenkins.perf.lab.eng.bos.redhat.com/job/pbench-ci-multi/job/main/110/)), Mr. Jenkins is no longer finding the Cobertura coverage report which the PyTest run creates.  (This corresponds to Git hash a53ddeb, which I believe is PR #2349.)  What's more, while the file seems to be created properly by the PyTest run in the container, when the `post` section of the pipeline is executed, the file is gone!

The previously-merged PR #2424 made changes which are relevant to this functionality.  (Its build, [#109](http://jenkins.perf.lab.eng.bos.redhat.com/job/pbench-ci-multi/job/main/109/), _failed_ due to non-deterministic behavior of one of the tests; otherwise, I believe it would have shown this behavior as well.)

The answer turns out to be that, while the file is created inside the container, the execution of  the `post` section is outside the container, and so the file systems are different, and the file exists but is not where we expect to find it.  Prior to PR #2424, `jenkins/run` was mapping the Jenkins temporary directory into the container as `/tmp`, which allowed the `post` section to find the file.  PR #2424 simplified the invocation of the container, and this detail was lost.

This commit changes the location of the coverage report to place it back into a mapped file system, and it puts it in a location which is easily accessible both inside and outside the container...and now things work again.  (And, it replaces the slightly confusing comment in the `Pipeline.gy` file with a new slightly less confusing comment.)
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant