Closed addyess closed 2 years ago
I wonder why this has suddenly started popping up (I've talked to three people in the last week that hit this). If it is a teardown problem, it feels like the recent changes to pytest-operator shouldn't have made it worse, so maybe there's also been a recent change to juju that has caused the increased rate of problems?
Nevermind ignore my previous comment - pinning to <0.17 fixes things so ...
the biggest change in this process is that the ops_test.plugin.forget_model
now waits for a model to tear down rather than just issue the command to "Destroy the model" without the waiting. I think there's an exception being raised in the waiting for teardown
code that isn't being caught which results in test errors in the clean up of ops_test
This other PR (https://github.com/charmed-kubernetes/kubernetes-metrics-server-operator/pull/1) is using this branch of pytest-operator to test the waters
@ca-scribner I've seen this seems to resolve the test teardown issues. Mind a review?
Seems to be working in this test PR that was originally hitting the same issue
Attempting to address https://github.com/charmed-kubernetes/pytest-operator/issues/72