Allowing replicated server to start
-
We've converted one of our hypervisors to XCPNG this weekend with new storage. I then setup CR to replicate one of our primary VM's across to the new server. The VM is about 2TB in size. I then followed the following steps:
- I shutdown the virtual server on the origin host
- I renamed the VM on the new server
- I removed the tag that said "continuous replication" - I believe this is added by default when CR is setup
- I started the VM on the new server. I got a warning saying that the VM cannot be started but I opted to force start as the CR is now "complete" and the server was already shutdown on the origin server
What I failed to do however was to delete the backup job, and simply be luck noted this morning the storage is 95% full and it was now busy replicating the shutdown VM accross to the storage
I restarted the toolstack on both servers and thankfully the backup jobs stopped and on the new server the "pending" vhd files was deleted and storage is resolved.If anyone ever uses CR as a method to migrate large VM's from 1 server to another, once migrated, DO NOT FORGET TO DELETE THE BACKUP JOB.
Over to the question, now that I removed the tag, renamed the VM and deleted the backup job, will the warning that that VM cannot start go away automatically or is there another setting that I need to set to tell XO that this is now the active VM?
-
This is exactly what we did for our own migration from Intel to AMD EPYC servers. That's often the best option to minimize downtime when live migration isn't possible. We call it "warm migration" here
Anyway, back to your question: as soon you force started the VM in Xen Orchestra, this will remove the warning on start for ever.
-
@olivierlambert you guys rock, thank you for all your hard work! Noted regarding the warning.
What was strange this time around was we've actually warm migrated another VM to new storage a month or so ago, and after the migration I shutdown the origin VM and attempted to do a last CR to ensure I get all of the data across. I however recalled that the backup did not want to run and I assumed it was probably because a shutdown VM cannot be CR. The backup job however of yesterday did indeed copy across the shutdown VM (or attempted to). Is there such a restriction that does not allow CR on a shutdown VM? We're doing another warm migrate in 2 weeks time and would obviously prefer to ensure all data is copied across as appose to a window period of data loss.
-
There's no issue nor restriction to replicate a shutdown VM.
CR will work as long as the destination VM never booted (otherwise destination blocks will be modified, so you understand now why we should block boot except if you really know what you do
)
