Closed DnPlas closed 1 month ago
lgtm except for one comment
Do we expect the integration CI to fail here?
So far yes, I have added a comment here. I am experimenting some solutions to unblock us.
I believe the CI was failing due to a mismatch in the versions that were deployed in these integration tests. This PR is attempting to bump the juju version, which also requires most packages to have a newer ops
version to keep compatibility. For one reason or another, the versions of the charms that get deployed as dependencies of the testing framework were not updated, causing the issue we observed (for details, please refer to https://github.com/canonical/bundle-kubeflow/issues/859#issuecomment-2034468799).
If everything goes well, #405 should fix the CI issues in this PR.
Closing as changes in this PR were introduced in #407 and will be introduced in #400
Bumping juju and ops packages to use them in newer versions of the charms, plus testing them in a CI with a more recent juju version.
This commit also skips some test cases that will be removed in a follow up commit introduced by canonical/istio-operators#401.
Part of canonical/bundle-kubeflow#859 Part of #398
Manual testing
juju config istio-pilot default-gateway="some-name"
andjuju relate istio-pilot istio-ingressgateway
track/1.17