You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
It would be interesting if this tool could verify if a charm with a set of alert rules is actually exposing the metrics those alert rules refer to.
This can be specially useful in the following cases:
If the metric name changes
If the metric is removed entirely, but the charm author did not remove the alert rule
If the metric even exists
The text was updated successfully, but these errors were encountered:
Agreed. cos-tool would then need to be able to extract metric names from alert rules.
Just realized there could be some minor edge cases that may affect testing:
When it comes to labels, some instrumentations have conditional metrics, e.g. api latency per path, and if the particular api endpoint wasn't probed yet, then the metric with the corresponding "path" label wouldn't exist.
Also related are aggregation operators: it would be difficult to make test assertion that the labels used by aggregation operators actually exist.
Enhancement Proposal
It would be interesting if this tool could verify if a charm with a set of alert rules is actually exposing the metrics those alert rules refer to.
This can be specially useful in the following cases:
The text was updated successfully, but these errors were encountered: