Merge pull request #10446 from thotz/hpa-reconiliation-issue
docs: mention about reconciliation issue in hpa docs
This commit is contained in:
1 file changed
+3
@@ -300,3 +300,6 @@ spec:
|
||||
sum(rate(ceph_rgw_put[2m])) # prometheus query used for autoscaling
|
||||
threshold: "90"
|
||||
```
|
||||
|
||||
!!! warning
|
||||
During reconciliation of a `CephObjectStore`, the Rook Operator will reset the replica count for RGW which was set by horizontal pod scaler. The horizontal pod autoscaler will change the again once it re-evaluates the rule. This can result in a performance hiccup of several seconds after a reconciliation. This is briefly discussed (here)[https://github.com/rook/rook/issues/10001]
|
||||
Reference in new issue
Block a user