HA trouble with SDN

[felie]

Hi,
I have a 3 PVE cluster (v9.2.4) with VM connected to SDN Vnets. Live VM migration is OK for VM without HA or VM in HA not using SDN VNets.
For VM in HA with VNets, migration is not permited (it was OK with linux bridges) :

Requesting HA migration for VM 201 to node srv-pve2
cannot migrate resource 'vm:201' to node 'srv-pve2':
- resource 'vm:201' not allowed on target node 'srv-pve2'
TASK ERROR: command 'ha-manager migrate vm:201 srv-pve2' failed: exit code 2

ha-manager status is OK ; FRR is Operational ;

Any idea of the trouble ?

 

[fba]

Hi,

I would check that:

• the SDN zone isn't limited to specific nodes and the target node is excluded (s. /etc/pve/sdn/zones.cfg)
• there is no ha-rule for restricting the nodes for this vm and the strict setting being enabled. (s. /etc/pve/ha/rules.cfg)

 

[felie]

Thank-you for answer. All nodes (no restriction) are selected in the zone and the PVE cluster are indicated in rules

node-affinity: ha-rule-c9f96d5d-8d26
        comment vanallen
        nodes srv-pve1:3,srv-pve2:2,srv-pve3:1
        resources vm:201
        strict 0

By the way, I made some new tests and I see that migration is OK if Affinity Rule for a VM is disable. So it seems that there is something wrong with Affinity Rules and VNets (for VM with same affinity rule bur connected to linux bridge, there is no trouble). Any idea ?

 

[felie]

[Solved] : for this proxmox version (may be also for somes versions before) for affinity rules, nodes may have same weight to permit live migration. My VM on srv-pve1 refused to move to srv-pve2 because "srv-pve1:3,srv-pve2:2,srv-pve3:1". If "srv-pve1:2,srv-pve2:2,srv-pve3:1" for exemple, migration to srv-pve2 is OK. No matter with SDN.

 

[dakralex]

To clarify this here, this is the expected behavior for the given HA affinity rule set. Essentially, if the HA resource has failback enabled (which it has by default) and is part of a node affinity rule with node priority classes, the HA resource can only be placed on nodes from the highest available priority class. Therefore, a recent change has disallowed these types of migrations to prevent them as they would be reverted soon anyway.

For further information, see one of these posts [0] [1].

[0] https://forum.proxmox.com/threads/a...-i-cannot-more-migrate-vm.185265/#post-862828
[1] https://forum.proxmox.com/threads/c...esource-vm-is-not-allowed.184806/#post-862830