This repository has been archived by the owner on Feb 7, 2024. It is now read-only.
This repository has been archived by the owner on Feb 7, 2024. It is now read-only.
Spurious numbers in metrics #430
Open
Description
In the 48 hours following the deployment of the Prometheus metrics endpoint, at least two bugs have been made apparent thanks to the Grafana dashboard:
- Failed tasks often (but not always) seem to be counted twice in
retrace_tasks_finished{result="fail"}. - The number of running tasks (
retrace_tasks_running) sporadically jumps up to wild numbers, such as 70, 18 or 39, for a few minutes at a time. The maximum allowed number of running tasks (MaxParallelTasks) is 12 on retrace.fp.org, so these numbers make no sense.
Activity