MDBF-991 - Obtain failed test results and the associated output from MTR - #816
Conversation
a638aa9 to
e3fabbd
Compare
|
Looks good! The syntax in bash for comparing integer should be I am wondering if we should not use Finally, you should probably have an error function since error should go to STDERR, see https://github.com/MariaDB/buildbot/blob/dev/scripts/bash_lib.sh#L37-L41. Those are not mandatory changes of course but nice to have I think. |
d2d9c90 to
68ceebd
Compare
|
Thanks @fauust for the review! See: 68ceebd PS: I cannot source the functions because the code is injected into a python string. The advantage is that the scripts are very portable, the disadvantage is obvious. |
Sometimes is useful, for non-critical buildbot steps, to mark them as warnings if they fail. We don't know in advance what is the non-zero return code of a script that marks a failure so is better to treat the 1-255 range as a Warning and only 0 - Success.
do not use interpolate for load_script. The only way variables can be passed to scripts is through command line arguments.
Standalone build step that checks for xml's files in the mtr log path and reports them back to the collector service. The collector service is running on the buildmaster as a docker container and its address is defined in the .env files This is not a critical client service so just warn on step failure. Common failures that can occur are: collector service is not available or the report is not a valid xml. If all MTR steps are successful then naturally there's no failure to report from the XML files so skip this step -> hasFailed(step)
68ceebd to
616a9bb
Compare
Client side implementation of the MTR reporter.
Follow along commit messages for implementation details.