CI: Making benchmark errors easier to find #29907
Conversation
datapythonista
added
CI
|
Looks like the desired error messages did show up, not sure if there are some non-desired ones in there too |
|
Cleaned it a bit, errors were being duplicated, not sure why. The log when not failures were found was not being shown. Not sure why I did that, but I think it makes sense that it's shown, so restored it. |
|
Sounds good. Some unwanted errors might go away after #29906. |
|
@jbrockmendel this looks good to me now. If you can have a look, I'll remove the failure once you're happy. Error can be seen here: https://github.com/pandas-dev/pandas/pull/29907/checks?check_run_id=324562585#step:13:3037 |
|
nice! |
| if [[ $(echo "$ASV_OUTPUT" | grep "failed") ]]; then | ||
| echo "##vso[task.logissue type=error]Benchmarks run with errors" | ||
| echo "$ASV_OUTPUT" | ||
| asv dev | sed "/failed$/ s/^/##[error]/" | tee benchmarks.log |
There was a problem hiding this comment.
Is there a reason for using tee here instead of simply redirecting the sed output?
There was a problem hiding this comment.
I want the log to be shown in the CI, tee gets it printed (besides saving the file later used to see if there are failures and to save the artifact)
|
ok lets try it out |
It takes a while to find the failing line of the benchmarks logs at the moment. This should highlight the failing ones, and also publish the log as an artifact so it can be analyzed easier.
CC: @jbrockmendel