Details
-
Improvement
-
Resolution: Fixed
-
Major
-
master
-
0
Description
Issues have been seen where it appears KV fails to serve a Prometheus scrape request within ~10s.
This has generally correlated with other activities being slow (e.g., ConnManager). It would be useful to also know how long Prometheus requests took, both overall and per-bucket.
It may also be valuable to add Slow Op style logging in this situation too.