-
Bug
-
Resolution: Unresolved
-
Normal
-
None
-
Quality / Stability / Reliability
-
3
-
False
-
-
False
-
-
-
Moderate
-
None
Description of problem:
In the multiclusterStatusAggregation_controller.go second manifestwork for loop, it gets status from all manifestwork for the same cluster with different timestamps. The manifestworks is not sorted by timestamp so if the last manifestwork has an older timestamp then the status will be outdated. We need to save the timestamp in appSetClusterStatusMap, compare the Manifest timestamp and only set the status if the timestamp is later than the current one.
Example:
feng-mort-pull-feng-mc3-1ba4a - Dec 13, 2024, 10:41 AM
feng-mort-pull-feng-mc3-39cce - Dec 17, 2024, 9:44 AM
feng-mort-pull-feng-mc3-89ea8 - Dec 18, 2024, 9:40 AM
feng-mort-pull-feng-mc3-d335e - Dec 16, 2024, 9:36 AM
The manifestwork may be listed in this order so at the end of the loop feng-mort-pull-feng-mc3-d335e's status will be used but from the timestamp feng-mort-pull-feng-mc3-89ea8 has the latest status.
It will be very hard for QE to verify this since there are many variables like the status of the Argo Application and the order in which the ManifestWork is listed. So this might be only verifiable by dev.
Version-Release number of selected component (if applicable):
How reproducible:
Steps to Reproduce:
- ...
Actual results:
Expected results:
Additional info:
- depends on
-
ACM-16187 ACM 2.11 Pull-Model does not reflect changes in ClusterReplicas
-
- Backlog
-