In my case, this is running against a `stable/zed` and it is almost _always_ reproducible. In order to save on memory and resources in our CI AIO, we tune down the number of workers to 2:
What is the number of workers that you are using in your CI? I wonder if the simple reproducer could be to set workers down to 2 and then just spam add/remove router interface..
Hi Miro,
In my case, this is running against a `stable/zed` and it is almost _always_ reproducible. In order to save on memory and resources in our CI AIO, we tune down the number of workers to 2:
https:/ /github. com/vexxhost/ atmosphere/ blob/stable/ zed/molecule/ aio/group_ vars/all/ molecule. yml#L159- L162
It almost always fails (fails more than it works) in that scenario:
https:/ /ci.atmosphere. dev/t/atmospher e/builds? job_name= atmosphere- molecule- aio-ovn& project= vexxhost/ atmosphere
What is the number of workers that you are using in your CI? I wonder if the simple reproducer could be to set workers down to 2 and then just spam add/remove router interface..