Closed rgildein closed 4 weeks ago
Thank you for reporting us your feedback!
The internal ticket has been created: https://warthogs.atlassian.net/browse/KF-5896.
This message was autogenerated
As @rgildein mentioned in the description of the issue, this error will go away if we remove these lines from the code.
I have a few comments on the overall design of this method, though.
deploy_and_assert_grafana_agent
will deploy the grafana-agent-k8s
charm into the testing model, then depending on the charm, some integrations will be added (metrics, dashboard, logging), and finally we will check if the grafana-agent-k8s
unit goes into BlockedStatus
. We expect it to go to this status because we are not integrating it with any of the "consumers", that is, none of the COS charms like loki or grafana are deployed and integrated in our testing framework, so by design grafana-agent-k8s
will just block (hence the logging-consumer: off, grafana-cloud-config: off
message).
Across the cos_integration
code, the grafana-agent-k8s
charm is only used for curl
ing the metrics endpoint of the charm under test.
It looks to me that:
curl
ing from the unit, the grafana-agent-k8s
charm is not used at all, we are just deploying and integrating it with the charm under test, but not actual functionality is used during the execution of the test.charm unit going into BlockedStatus based on the knowledge we have about the charm. It is a not the best practice to rely on a unit status other than
Active` as it can be affected by different scenarios and it is prone to change (as we saw with this issue).Re-opening as we still need to bump chisme versions in repos where this code is used.
Chisme 0.4.1 is out, and is now updated in all repos. We can close this task.
Bug Description
The charm grafana-agent-k8s was moved back to revision 45 in stable channel. (e.g. this action shown that rev 68 was used before) This cause an issue, since status message is different from it was expected.
After some discussion in team, we think the abstraction she should not obey the workload status message after grafana-agent-k8s is deployed. It means that these lines should be removed.
Currently, the workaround is to deploy grafana-agent-k8s from latest/edge channel , e.g.
deploy_and_assert_grafana_agent(..., channel="latest/edge", ...)
To Reproduce
tox -e integration
Environment
Juju 3.4+ Microk8s 1.24+
Relevant Log Output
Additional Context
No response