-
Bug
-
Resolution: Unresolved
-
Blocker
-
CNV v4.19.0
-
Quality / Stability / Reliability
-
0.42
-
False
-
-
False
-
None
-
-
None
Description of problem:
When calling eviction on a VM pod, then kubevirt catches it, and starts to trigger LM of VM in order to evict the pod eventually. If LMs fail, then Kubevirt fails to LM the VM, and as such the eviction fails. However, the annotation set on the VM pod returns that the eviction was successful
Version-Release number of selected component (if applicable):
4.19
How reproducible:
Always
Steps to Reproduce:
1. Evict a VM pod 2. Make the LM fail 3.
Actual results:
KubeVirt will retry LM Evict return code will indicate success already
Expected results:
KubeVirt will retry LM Evict return code does not indicate success, but in progress
Additional info:
Discussed in https://groups.google.com/g/kubevirt-dev/c/wtFCrxwj4SE
- blocks
-
CNV-58437 [GA] CPU Load Aware rebalancing with Descheduler
-
- In Progress
-
- is duplicated by
-
WRKLDS-1721 Descheduler triggers new evictions while virt-controller retries failed migrations
-
- Closed
-
- links to