Some IPs do not failover when entering Persistent CARP Maintenance Mode
markyochoa last edited by
It looks like it's just the first 20 IPs that don't fail over. Anyone know why this is?
When it goes into maintenance mode, they all should fail over. There isn't any logic in there that I'm aware of that would make it stop at a particular number.
That said, if there is no other system on the network that is capable of taking over as MASTER, then maintenance mode will not necessarily show BACKUP for all VIPs. It sets it up so that it will demote itself below others, but if nothing else takes over, it will still be MASTER.
ipodgorny last edited by
I'm actually seeing the same thing. 3 ot of 5 fail over while 2 stay as master. What weird is 2nd node becomes a Master on all 5 as well most of the time, and sometimes stays as backup.
This creates whole a lot of issues. I've also noticed that OpenVPN doesn't failover when CARP IP is used, maybe it is related. Service fails to register IP and exits with fatal error, and can't be started withour rebooting appliance even when CARP is disabled on the first node and second is definetly a master. Service crashes completelly on a first node when CARP is re-enabled and it becomes a master again.
cmb last edited by
Maintenance mode just bumps the skew to 254. That means it's backup status only if it sees advertisements from a lower skew/higher priority. Absent that, it's still master.
I'm guessing in that case the ones that don't go to backup status are on one particular interface. Likely CARP advertisements don't make it from secondary to primary on that interface for some reason. Most always network-related, either no connectivity between them, or multicast not making it in that direction.