-
Bug
-
Resolution: Done
-
Blocker
-
openshift-4.11.z
-
None
-
False
-
None
-
False
-
-
-
-
Hypershift Sprint 20
-
Critical
-
10
-
0
-
0
During PerfScale testing of 80 Hosted Clusters on 24 m6i workers we ran into an issue where OpenShift reports the max allocate-able EBS volumes to be 39 (via oc describe node) however anything above 27 attached EBS volumes causes the attachment to never complete and sit with a status of "attaching."
NOTE: both m6i and m6a instance types reported this as 39
On m5 instance types this number is set to 25 and thus does not run into this issue. Ultimately, this caused many ovnkube pods to be unable to start as their pvs were not present and they timed out.
Example pod output:
Warning FailedAttachVolume 96s (x27 over 40m) attachdetach-controller AttachVolume.Attach failed for volume "pvc-763be113-1923-4462-92b5-62f64517578d" : rpc error: code = Internal desc = Could not attach volume "vol-03441aff5f07550bd" to node "i-09190d2d740bf2212": attachment of disk "vol-03441aff5f07550bd" failed, expected device to be attached but was attaching
- is blocked by
-
OCPBUGS-3775 attachable-volumes-aws-ebs incorrect for AWS m* instance types
- Closed