I am looking to upgrade my cluster communication/DRBD from 1GB connections to 10GB connections. Does anyone have any suggestions on the best way to accomplish this without losing anything important?
Here is what I am thinking.
1. Move VM's to node 2
1a. Install 10GB card
2. Move VM to node 1
2a. Install 10GB card
3. Setup Networking
4. Remove node 2 from the cluster
5. Add node 2 back to the cluster using the 10GB adapter address
6. Check cluster (pvecm status/pvecm nodes)
7. Edit /etc/drbd.d/r0.res and /etc/drbd.d/r1.res to reflect new network
8. Restart DRBD
I am thinking DRBD should be as simple as just changing the config files and issuing "drbdadm adjust all". Hopefully this doesn't cause a split brain, I am guessing it would be best to have all VM's on 1 node at all times during the upgrade just in case.
I'm also a bit confused on setting cluster communication back up. If I simply remove node 2 from the cluster, then add it back in using the new 10GB network does anything have to be done with node 1? Looks like I need to change /etc/hosts on each node to match the new interface. Can anyone confirm, does cluster communication need to be on eth0/vmbr0, or can it be on any adapter (eth7/vmbr7)?
Here is what I am thinking.
1. Move VM's to node 2
1a. Install 10GB card
2. Move VM to node 1
2a. Install 10GB card
3. Setup Networking
4. Remove node 2 from the cluster
5. Add node 2 back to the cluster using the 10GB adapter address
6. Check cluster (pvecm status/pvecm nodes)
7. Edit /etc/drbd.d/r0.res and /etc/drbd.d/r1.res to reflect new network
8. Restart DRBD
I am thinking DRBD should be as simple as just changing the config files and issuing "drbdadm adjust all". Hopefully this doesn't cause a split brain, I am guessing it would be best to have all VM's on 1 node at all times during the upgrade just in case.
I'm also a bit confused on setting cluster communication back up. If I simply remove node 2 from the cluster, then add it back in using the new 10GB network does anything have to be done with node 1? Looks like I need to change /etc/hosts on each node to match the new interface. Can anyone confirm, does cluster communication need to be on eth0/vmbr0, or can it be on any adapter (eth7/vmbr7)?
Last edited: