Replacing all drives on clustered nodes

[kriev98]

Hey,

I have 2 nodes (identical config) in a cluster with HA (QDevice for 3rd vote) on a ZFS Pool. I need to upgrade all drives (we are doubling the capacity). How should i approach this. All VMs are backed up to a PBS server as well.

Is it possible to move everything to NodeB, disable HA on VMs (force them to stay on NodeB) delete ZFS Pool on NodeA, replace drives, recreate pool on NodeA, reanable HA, push everything over NodeA and repeat on NodeB?

 

[mariol]

Provided that the RAID level remains the same, you can also replace all drives while the system is under normal state. Enable “Autoexpand” [0] to automatically expand each pool after replacing all drives.


[0] https://docs.oracle.com/cd/E19253-01/819-5461/githb/index.html

 

[kriev98]

The hardware is in a datacenter, i will not be doing the manipulations nor will i have contact with the person doing it. So the server will go down and once back up, the new drives will be in (8 bay server, 2 NVMes (boot drives, won't be touched) and 6 SSDs, no hardware raid, will be replaced at once). I just need to make sure how to prep the system before that shutdown to just recreate the ZFS pool with the same name. The only issue i see is that the pool on NodeA will be double the size than NodeB, will that be an issue when reenabling HA and bringing back all VMs to NodeA?

 

[UdoB]

The hardware is in a datacenter,

What I would do in beforehand is to setup a virtual cluster with similar topology (but much smaller) devices. Then one can simulate/train such a delicate operation.

The bad news: to create such a thing may be a complex task in itself. But either a virtual test-cluster or a real hardware test-cluster is really useful and makes you sleep better because you can train complex tasks.

Of course: ymmv ;-)