Action running with timeout
Xuanwo opened this issue · comments
Sometimes this action will exit with error timeout:
https://github.com/Xuanwo/Xuanwo/runs/4790266447?check_suite_focus=true
Related to #3 (comment)
I made a change to and export the trace.zip even it timeout a622d7f
OK, I will attach the trace.zip
here if I meet timeout again.
I try to re-run multiple times https://github.com/pengx17/knowledge-garden/actions/workflows/main.yml
Seems it did not timeout so far 😅.
Reproduced: https://github.com/Xuanwo/Xuanwo/runs/4799772733?check_suite_focus=true
Oh, but this action runs using old versions..
Wired, it seems trace.zip
not generated correctly: https://github.com/Xuanwo/Xuanwo/runs/4813142697?check_suite_focus=true
Wired, it seems
trace.zip
not generated correctly: Xuanwo/Xuanwo/runs/4813142697?check_suite_focus=true
Seems I forget to rebuild the docker image 🙃
Maybe I start the graph loading too early?
Tried a fix in cdce58f, and as a result, publish action has been working properly for 7 hours (28 runs).
The issue might be that Logseq cannot load a graph too early when app is initializing. Should not be a problem in the real world, but could be an issue for automation.
Maybe we need help for logseq team. For example, can they trigger an event after app initialized / load finished / load succeed?
Usually, for UI testing, it is a common best practice to depend on visual UI elements, like text, icons, aria labels etc. The fix in cdce58f did try to testify if the loading spinner is still shown to see if the app is still initializing.
The load success event in Logseq is a non-platform-agonistic native dialog, but It is not intercepted/automatable by Playwright yet. I think we could replace it with a browser dialog instead.
I think the issue has been fixed. I will close it for now. Feel free to reopen it or create a new one.