I stand corrected. Just took a minute to sync :swe...
# harvesterp
powerful-easter-15334
05/21/2025, 10:43 AMI stand corrected. Just took a minute to sync 😅
👍 1
h
happy-cat-90847
05/21/2025, 11:17 AMI think this needs to be improved too. There is a sync ratio. How often do these backups happen? If so, then there is a setting from Longhorn that can check with a matching frequency. I expect us to provide a way to deliver DR volumes in which the VM spec is also syncd. But the work really is in the fail-back and setting back the failover site to sync back to the primary.
p
powerful-easter-15334
05/21/2025, 11:20 AMBackups should happen every day (though right now they're not because of another issue XD) but I don't think I'll run a restore every day. First because it's manual work and also because of said other issue.
I was keeping track of the DR feature request on the Harvester GH. I believe it went from planning to 1.4.0 to 1.5.0 to 1.6.0 and back to planning 😆
I have a few ideas on that concept though.
powerful-easter-15334
05/21/2025, 11:25 AMI was thinking for the DR-like feature, it could be:
• Initial VM backup & restore as normal
• Volumes of DR VM in DR state (using Longhorn's DR feature). This way the volumes stay as up-to-date as possible. Shouldn't be an issue as the DR VM would stay off anyway.
• When the DR VM is activated, the volumes are moved out of DR state and are used.
When the DR has served its purpose and it's time to return to prod:
• DR VM is backed up
• Prod restores the backup into the existing (but outdated) VM
• Prod VM can restart running
• DR VM (which is not in DR mode) gets destroyed and step 1 is run again. I say this because Longhorn doesn't let you turn an activated DR vol back into DR