Open npriebe opened 1 year ago
Same for me, with the same spi and keycloak versions.
@npriebe adding username to the metric would increase cardinality and afaik it is generally considered poor practice to do that in TSDBs
e.g.
https://grafana.com/blog/2022/02/15/what-are-cardinality-spikes-and-why-do-they-matter
@npriebe did you find any solution ? I find the values as wrong since in every metrics, it's showing me large values (for ex. keycloak_logins_created{realm="xyz",provider="keycloak",client_id="service",} 1.688189803383E9
which converts to 1688189803.383
when displayed in prometheus/grafana
) for all the clients which doesn't make any sense (considering the amount of requests stats)
Description
Hi, I noticed that for certain metrics either labels are missing or the labels give wrong values.
The following metrics still need labels:
keycloak_logins, keycloak_failed_attempts, keycloak_client_logins, keycloak_failed_login_attempts:
With this metric, it was useful to have the username as a label for understanding where the request is being attempted.
keycloak_response_errors:
With this metric, the realm as a label is missing.
The following metrics has wrong label values:
keycloak_failed_client_login_attempts:
Here has the label "client_id" wrong values. For the other metrics, the name of the client is set as client_id. Example:
Expected Behavior
Metrics with the described labels.
Actual Behavior
Labels seem to be missing.
Environment