CARP backup flapping to master and back randomly

cr_hyland

We have an installation on ESXi 5.5 on HP Blades with pfSense on 2x separate blades and using E1000 nics.

Have been using CARP for years with no issues.
Since we rebuilt (clean installs) both firewalls on 2.2 we have been experiencing CARP weirdness.

Several time throughout the day, for what seems like no reason we get mail notifications that our backup switched all VIPs to master and back to backup again all within a few seconds.
The master never complains or sends any mail notifications that it switched to backup for a moment.

Firewall logs don't show anything unusual, traffic graphs are normal. Limiters show no drops. Queues are normal. Gateway latency on both firewalls is perfect.
Interface statuses show no errors. State tables are all around 50k which is normal for our load and don't increase around the time of the CARP switch.
VM memory and CPU usage is all low and same goes for ESXi hosts.

Really strange, seems to be absolutely no reason for it.

michaelschefczyk

Please try adding net.inet.carp.senderr_demotion_factor=0 to tunables.

cr_hyland

@michaelschefczyk:

Please try adding net.inet.carp.senderr_demotion_factor=0 to tunables.

Hi, what does this do exactly?
Will it affect the functionality of my CARP failover?

gordc

Did you ever find an answer for this?

I have the same thing happening.