Migration and two networks [solved]

[rmoriz]

solved: https://bugzilla.proxmox.com/show_bug.cgi?id=1177#c2


I've the following setup:

- 2 PVE nodes:
pve1 has 2 interfaces:
- 10.0.1.15 1Gbit/s
- 10.0.40.10 2.5/Gbit/s

pve2 has 2 interfaces:
- 10.0.1.16 1Gbit/s
- 10.0.40.11 2.5/Gbit/s

- the 1G interfaces are considered for external services (VMs, CTs), the 2.5G exclusively for migration (inter-cluster communication).
- Nodes can connect to each other on the different networks.
- Both nodes form a Proxmox cluster, link 0 is the 2.5G IP/IF and link 1 the 1G IP.
- Corosync.conf excerpt:

nodelist {
  node {
    name: pve1
    nodeid: 2
    quorum_votes: 1
    ring0_addr: 10.0.40.10
    ring2_addr: 10.0.1.15
  }
  node {
    name: pve2
    nodeid: 1
    quorum_votes: 1
    ring0_addr: 10.0.40.11
    ring2_addr: 10.0.1.16
  }
}

Issue:

When I migrate a VM from one node to another, the slow 1Gbit/s interface is used.

For example I move VM 104 from pve1 to pve2:

/usr/bin/ssh -e none \
-o BatchMode=yes \
-o HostKeyAlias=pve2 \
-o UserKnownHostsFile=/etc/pve/nodes/pve2/ssh_known_hosts \
-o GlobalKnownHostsFile=none root@10.0.1.16 \
-o ExitOnForwardFailure=yes \
-L /run/qemu-server/104_nbd.migrate:/run/qemu-server/104_nbd.migrate \
-L /run/qemu-server/104.migrate:/run/qemu-server/104.migrate \
/usr/sbin/qm mtunnel

I expected that the connection is made to 10.0.40.11 and not 10.0.1.16. What did I miss?

Thank you!
Roland