-
Bug
-
Resolution: Unresolved
-
Major
-
None
-
4.18
-
Quality / Stability / Reliability
-
False
-
-
None
-
None
-
None
-
None
-
None
-
CNF RAN Sprint 275
-
1
-
None
-
None
-
None
-
None
-
None
-
None
-
None
Description of problem:
When a CGU is blocked by another CGU, TALM will perform the "afterCompletion" actions while in the blocked state. The actions should not execute until the CGU has completed.
Version-Release number of selected component (if applicable):
CSV: topology-aware-lifecycle-manager.v4.18.0 registry.redhat.io/openshift4/topology-aware-lifecycle-manager-rhel9-operator@sha256:bd2bf0ef13ae56a483ac80d5979dabe6a12e6675888044edda419a063b7fcf64
How reproducible:
100%
Steps to Reproduce:
1. Create 2 CGUs. The second has an afterCompletion action to delete a label. The second CGU should be blocked on the first.
apiVersion: ran.openshift.io/v1alpha1
kind: ClusterGroupUpgrade
metadata:
name: upgrade-18
namespace: default
spec:
actions:
afterCompletion:
deleteClusterLabels:
upgrade-version-4-18: ""
deleteObjects: true
beforeEnable:
deleteClusterLabels:
upgrade-version-4-17: ""
backup: false
clusters:
- cnfdf02
enable: true
managedPolicies:
- core-upgrade-prep-18
- core-upgrade-ocp-18
- core-upgrade-olm-18
- core-upgrade-validate-18
preCaching: false
preCachingConfigRef: {}
remediationStrategy:
maxConcurrency: 1
timeout: 240
---
apiVersion: ran.openshift.io/v1alpha1
kind: ClusterGroupUpgrade
metadata:
name: upgrade-finish
namespace: default
spec:
actions:
afterCompletion:
deleteClusterLabels:
upgrade-finish: ""
deleteObjects: true
backup: false
clusters:
- cnfdf02
enable: true
managedPolicies:
# These policies unpause the MCP
- core-upgrade-workers-1-storage
- core-upgrade-workers-2
- core-upgrade-workers-3
preCaching: false
preCachingConfigRef: {}
remediationStrategy:
maxConcurrency: 1
timeout: 240
blockingCRs:
- name: upgrade-18
namespace: default
2. Apply the first CGU, then the second. The first should take some time to complete.
3. Note that the label is removed from the managedcluster while the second CGU remains blocked
Actual results:
ManagedCluster has the "upgrade-finish" label removed as soon as I add it and immediately if I re-add it.
Expected results:
Label remains on cluster until the second CGU completes.
Additional info:
- blocks
-
-
- Verified
-
- is cloned by
-
-
- Verified
-
- links to