API performance consistency
I'm using Gitlab standalone setup on pretty powerful server (16 cores, lots of RAM). Recently created a tiny service utilizing Gitlab API for internal usage. After a few days of using it, I realized about 10% of requests take more than a second to process, however rest of them are getting processed within 100ms.
API calls I'm using are the same all the time:
- list projects within a group
- show particular file in repo
Apache benchmark results:
Percentage of the requests served within a certain time (ms)
50% 79
66% 86
75% 90
80% 96
90% 121
95% 1787
98% 1908
99% 1912
100% 2006 (longest request)
Is there a way to make API requests processing time more consistent?