Cache charmcraft pack container, skip unstable tests except on schedule #95
Conversation
carlcsaposs-canonical
requested review from
taurus-forever,
dragomirp and
marceloneppel
This reverts commit ae78917.
| on: | ||
| pull_request: | ||
| paths-ignore: | ||
| - '.gitignore' | ||
| - '.jujuignore' | ||
| - 'LICENSE' | ||
| - '**.md' | ||
| schedule: | ||
| - cron: '53 0 * * *' # Daily at 00:53 UTC |
There was a problem hiding this comment.
Not an issue with the PR itself, but we may consider running the scheduled tests during slightly different times to not stampede on the runners.
There was a problem hiding this comment.
For now, I think it's okay to stampede the runners overnight—the runners will just queue up & run as soon as runners are available
There was a problem hiding this comment.
Starting all MySQL and PostgreSQL repos simultaneously at 00:53 might be noticeable in Los Angeles (16:53 local time).
There was a problem hiding this comment.
Starting them later might be noticeable in India, and I figured it'd be better to have a queue at the end of the day instead of the beginning of the day
| from pytest_operator.plugin import OpsTest | ||
|
|
||
|
|
||
| @pytest.fixture(scope="module") |
There was a problem hiding this comment.
I think we can even go with session here, it should never change during execution.
There was a problem hiding this comment.
Technically I think that's correct.
However, it multiple modules are run in a test, I think different instances of ops_test should be created since ops_test has attributes for the test model. (Different modules create different models)
In the mysql charms, I left the scope to the default of function to avoid changing the behavior of pytest-operator. The only reason to use scope="module" is to be compatible with other fixtures in the postgres charms. I'm not sure if this is a solid approach, but it seems like it's okay. It does feel a bit hacky already though
Ported from canonical/mysql-k8s-operator#146