Closed jelly closed 5 days ago
Touching machineslib.py
does not make tests re-run 3 times. Lets touch one test file for science.
@martinpitt please take a look, as changing only machineslib.py
does not trigger re-tries I still have a debug commit so you can see that the change did in fact make CI less flaky :)
@martinpitt requested some investigation if this is due to too many parallel runs, soon we have --amplify
to easily check this
@jelly If waiting longer helps, by all means go for it! It's plausible if that is a "soft" shutdown, i.e. the OS actually has to shut down (as opposed to just killing the VM). It's all emulated, so the inner VMs are sloooow :snail:
@jelly If waiting longer helps, by all means go for it! It's plausible if that is a "soft" shutdown, i.e. the OS actually has to shut down (as opposed to just killing the VM). It's all emulated, so the inner VMs are sloooow 🐌
I did some more testing in https://github.com/cockpit-project/cockpit-machines/pull/1705 too much amplification makes the VM crash, like here. So I'd say this is good to merge?
The top flake for Fedora-40, basically all the failures are waiting on the host to shutdown. Let's just bump the timeout and see if it works.
https://cockpit-logs.us-east-1.linodeobjects.com/pull-1693-69328051-20240626-133906-fedora-40/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1697-547cf65d-20240626-154604-fedora-40-firefox/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1697-547cf65d-20240626-154604-fedora-40-firefox/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1679-576c973f-20240616-083805-fedora-40/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1679-576c973f-20240616-083805-fedora-40/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1697-547cf65d-20240626-144920-fedora-40-firefox/log.html https://cockpit-logs.us-east-1.linodeobjects.com/pull-1697-547cf65d-20240626-154604-fedora-40-firefox/log.html