Currently if we want to debug any failed test case in jenkins regression test build, the only thing we can do is trying to rerun it again and expect it can be reproduced.
To improve this debugging process, we'd like to directly investigate the environment of this test build (directly access this cluster and Longhorn to check any states we're interested in).
To achieve this, we need the following improvement:
Describe the items of the test development (DoD, definition of done) you'd like
[ ] Pause the test when there are something needed to be investigate, so we can keep the problematic environment instead of continuing running following test cases
[ ] Any approach to prevent this test build from being cleanup by the cleanup resources pipeline cron job
[ ] Expose the information about how to access this cluster (like k3s.yaml) to somewhere (s3 or jenkins)
[ ] Create node port of longhorn-ui for each jenkins regression test build. Currently our longhorn-setup.sh wouldn't expose longhorn-ui, but it's could be easier for us to have visualization when debugging
Additional context
Add any other context or screenshots about the test request here.
innobead
commented
1 year ago
What's the test to develop? Please describe
Currently if we want to debug any failed test case in jenkins regression test build, the only thing we can do is trying to rerun it again and expect it can be reproduced.
To improve this debugging process, we'd like to directly investigate the environment of this test build (directly access this cluster and Longhorn to check any states we're interested in).
To achieve this, we need the following improvement:
Describe the items of the test development (DoD, definition of done) you'd like
Pause the test when there are something needed to be investigate, so we can keep the problematic environment instead of continuing running following test casesAny approach to prevent this test build from being cleanup by the cleanup resources pipeline cron jobExpose the information about how to access this cluster (like k3s.yaml) to somewhere (s3 or jenkins)Create node port of longhorn-ui for each jenkins regression test build. Currently our longhorn-setup.sh wouldn't expose longhorn-ui, but it's could be easier for us to have visualization when debuggingAdditional context
Add any other context or screenshots about the test request here.