I have 3 nodes, but only two in storage replication in zfs, currently from pve02 to pve01, and pve03 only in lvm as a voting member, my cluster and my vms were created in pve02 on a server weaker than pve01 to make sure that the workload would work well, but as my pve01 server is a newer server, I would like to move my vms to it and leave the operation on it, if I go to the vm and do a migration on it can this somehow cause a problem? Is there any risk in transferring the vm to pve01 and leaving pve01 as a priority in my HA? remembering that my vms were created on pve02.
vm migration in storage replication
- Thread starter dev03jr
- Start date
-
- Tags
- replication cluster zfs
replication in zfs is already happening from pve02 to pve01 with the job in replication, I would just like to move the vms already created in pve02 to pve01, I wanted to confirm if clicking on the vm and going to migrate this process would work, and if there is any risk.
From the storage perspective, you can do both live and offline migration from pve02 to pve01. The replication job will automatically change direction from pve02->pve01 to pve01->pve02.
@_gabriel live migration with local storage is possible, simply will take much longer as it has to copy the disk(s) too. If using replication, it will have to run one las sync job only, so the migration will faster (most of the data has been replicated already).
There's something else to care for, related to the CPU configuration of the VM and the CPU of each PVE node. CPUs must be at least equivalent from the perspective of supporting all the instructions that the CPU configured in the VM will use.
Which CPU type have your configured in the VMs?
What's the output of
@_gabriel live migration with local storage is possible, simply will take much longer as it has to copy the disk(s) too. If using replication, it will have to run one las sync job only, so the migration will faster (most of the data has been replicated already).
There's something else to care for, related to the CPU configuration of the VM and the CPU of each PVE node. CPUs must be at least equivalent from the perspective of supporting all the instructions that the CPU configured in the VM will use.
Which CPU type have your configured in the VMs?
What's the output of
lscpu in pve01 and pve02?
Forgot to mention that you can create a test VM in pve02, add it to replication and test live migration before proceeding with your producction VMs.
my pve02 has a cpu (2x Intel xeon E5-2407 v2 - 2.40GHZ), and my pve01 has (2x- INTEL XEON BRONZE 3106 1.70GHZ), I left the configuration in the vm in (x86-64-v2-AES) fully scaling to the weakest server and currently it is serving well, but as pve01 is a newer server, I would like to leave the weakest as a backup in case pve01 fails, and I would like to know if in this situation there is any risk in migrating the vm to pve01 and leaving it working with reverse replication. (pve01 -> pve02)