We recently enabled job metrics, and it is hammering our gitlab instance to the point where we have had to shutdown GCPE.
The job metrics pull does not appear to support max_age or most_recent parameters, and as such, appears to be trawling through hundreds of pages of jobs, 100 at a time, for each project (~150).
Additionally, the job metrics doesn't appear to respect the limit of 1 rps to gitlab. We see it spike at first to over 600 requests per min (10+ RPS) then it falls back down to around 300 requests per min till the gitlab server starts to fall over, spewing 500's for a little while, then the 502's start when the health checks restart the nodes.
We are willing to believe it is a configuration issue on our end, but I did look at the code and couldn't see any option for max_age or most_recent options.
We recently enabled job metrics, and it is hammering our gitlab instance to the point where we have had to shutdown GCPE.
The job metrics pull does not appear to support
max_ageormost_recentparameters, and as such, appears to be trawling through hundreds of pages of jobs, 100 at a time, for each project (~150).Additionally, the job metrics doesn't appear to respect the limit of 1 rps to gitlab. We see it spike at first to over 600 requests per min (10+ RPS) then it falls back down to around 300 requests per min till the gitlab server starts to fall over, spewing 500's for a little while, then the 502's start when the health checks restart the nodes.
We are willing to believe it is a configuration issue on our end, but I did look at the code and couldn't see any option for max_age or most_recent options.