I suspect this work is
a) Adding gs://kubernetes-jenkins/pr-logs/pull to buckets.yaml
b) speeding up the whole process so it doesn't take a million years to complete
c) most likely by
gs://BASE/JOB containing a list of BUILD folders, you instead have a bunch of BUILD.txt files (example) that contain the name of the gs://OTHER_BASE/JOB/BUILD/ path that contains the actual data files we need to parse.@crimsonfaith91 this could be a good project as well
I will start working on this issue. :)
Note that it looks funky since we just got rid of merge-blocking jobs.
@crimsonfaith91 thanks! Sent you an invite to collaborate on this repo. Once you receive it you can /assign this issue to yourself.
@fejta noted! i will work on this issue first. thanks! :)
/assign @crimsonfaith91
@fejta Do you mean changing this when you mentioned BUILD.txt?
I apologize if it is not a good question. I am new to the Jenkins repo (and Python). Thanks! :)
Intent here is to note the differences:
logs/JOB_NAME/BUILD_NUMBER and in that folder is a bunch of files like started.json, finished.json, etc.pr-logs/directory/JOB_NAME path has a pr-logs/directory/JOB_NAME/BUILD_NUMBER.txt file, the contents of which are a gs://kubernetes-jenkins/pr-logs/PR_NUMBER/JOB_NAME/BUILD_NUMBER that store a link to the actual started.json etc filesSo we will need to make the gen_json.py file aware of this directory pattern -- read the poor man's symlink at BUILD_NUMBER.txt and read the actual path to get the json/junit files.
thank you very much for the very detailed explanation! many thanks!
@fejta I will get back to this when I resolve the inactive reviewer issue. Sorry for the delay!
Testgrid and triage sites have deprecated the 24 hr test history