-
Task
-
Resolution: Done
-
Undefined
-
ACM 2.11.0
-
None
It is useful to record a passive activation operation and be able to share it with all the hubs using the same OADP storage location
This information could be useful in many DR scenarios to make sure there is no data corruption, for example :
- hubA backs up data
- hubB is the passive hub
- hubA goes down unexpectedly
- hubB takes over - restores activation data. The user doesn't create yet a backup schedule on hubB
- a new hub, hubC is set now as passive - passive data is being restored every 10min
- at some point hubA recovers ; hubA backupSchedule is active now and since no other hub writes backups ( user had not created a backupSchedule on hubB ) , hubA continues to write backups. Hub C gets the backup from hubA instead of hubC
If we can record the restore of managed clusters on hubB, when hubA recovers and starts up, it will check :
- if there is a record a passive activation operation ( gets the timestamp ), if such record exists,
- take the more recent
- check the hubA backupSchedule timestamp against the recorded passive activation timestamp. If the passive activation is more recent ( and the passive activation was not run on hubA) , then hubA is no longer the active hub and should not create backupSchedule. The backupSchedule status should be changed to BackupCollision and the velero Schedule should be removed. ( as done in this scenario, when hubB has a backupSchedule running - https://access.redhat.com/documentation/en-us/red_hat_advanced_cluster_management_for_kubernetes/2.10/html/business_continuity/business-cont-overview#avoid-backup-collision )