Monitoring

[Paul2708]

Hey there,

I was wondering if there are plans to provide metrics (e.g., via Prometheus endpoint) or if you already use some monitoring/visualization (e.g., Grafana with grader-service database as a datasource) for your deployments.

[florian-jaeger]

We want/need this as well, but there are no implementation plans yet!
Do you have any specific metrics that you require/want?

[Paul2708]

Besides simple numerical values (e.g., number of lectures, assignments, submissions), I would be interested in:

• submissions plotted by time to see on which dates the grader service is used the most
• status (i.e., success, failure) and duration of autograding pods
• total size of all assignment and submission-related files (i.e., the Git repositories)

However, all of the above is "nice to have" and not a showstopper in any way. We are already using the logs to create alerts on errors, so it's more about getting insights about how the grader service is actually used, because most of the time, we are only notified if something went wrong