CARP master stays in BACKUP state indefinitely

[clarknova]

OPNsense 21.1.7_1-amd64

I have two pairs of firewalls and they're both behaving in the same way and I'm not sure why. The primary has CARP interfaces with base:skew values of 1:0. Secondary is 1:100. Occasionally the primary shows CARP status of BACKUP and secondary shows MASTER. I see this in the log on the primary while in BACKUP status (newest on top):

Code: [Select]

2021-07-06T15:10:26 opnsense[54731] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "192.168.1.254 - LAN CARP (1@em1)" has resumed the state "BACKUP" for vhid 1
2021-07-06T15:10:26 kernel em1: deletion failed: 3
2021-07-06T15:10:26 kernel carp: 1@em1: MASTER -> BACKUP (more frequent advertisement received)
2021-07-06T15:09:51 opnsense[23015] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "10.2.0.1 - LDC01-TEST CARP (2@em0_vlan2)" has resumed the state "BACKUP" for vhid 2
2021-07-06T15:09:50 opnsense[76639] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "10.100.2.1 - LDC01-TDC01-L2 (3@em0_vlan910)" has resumed the state "BACKUP" for vhid 3
2021-07-06T15:09:50 kernel em0_vlan2: deletion failed: 3
2021-07-06T15:09:50 kernel carp: 2@em0_vlan2: MASTER -> BACKUP (more frequent advertisement received)
2021-07-06T15:09:50 kernel em0_vlan910: deletion failed: 3
2021-07-06T15:09:50 kernel carp: 3@em0_vlan910: MASTER -> BACKUP (more frequent advertisement received)
Following the page at https://docs.opnsense.org/development/backend/carp.html, I see the following on the primary (while in BACKUP status):

Code: [Select]

root@LDC01A:~ # sysctl net.inet.carp.demotion
net.inet.carp.demotion: 1048576
root@LDC01A:~ # sysctl net.inet.carp.ifdown_demotion_factor
net.inet.carp.ifdown_demotion_factor: 240
root@LDC01A:~ # sysctl net.inet.carp.senderr_demotion_factor
net.inet.carp.senderr_demotion_factor: 240
root@LDC01A:~ # sysctl net.inet.carp_demotion_factor
sysctl: unknown oid 'net.inet.carp_demotion_factor'
root@LDC01A:~ # sysctl net.pfsync.carp_demotion_factor
net.pfsync.carp_demotion_factor: 240

Not sure what to make of it, I do this:

Code: [Select]

root@LDC01A:~ # configctl interface update carp service_status
OK
Within seconds, the primary firewall changes its CARP status to MASTER and the secondary to BACKUP. I see this in the log on the primary:

Code: [Select]

2021-07-07T10:44:02 opnsense[99061] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "10.2.0.1 - LDC01-TEST CARP (2@em0_vlan2)" has resumed the state "MASTER" for vhid 2
2021-07-07T10:44:01 opnsense[54887] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "192.168.1.254 - LAN CARP (1@em1)" has resumed the state "MASTER" for vhid 1
2021-07-07T10:44:00 opnsense[21941] /usr/local/etc/rc.syshook.d/carp/20-openvpn: Carp cluster member "10.100.2.1 - LDC01-TDC01-L2 (3@em0_vlan910)" has resumed the state "MASTER" for vhid 3
2021-07-07T10:43:59 kernel carp: 2@em0_vlan2: BACKUP -> MASTER (preempting a slower master)
2021-07-07T10:43:59 kernel carp: 1@em1: BACKUP -> MASTER (preempting a slower master)
2021-07-07T10:43:59 kernel carp: 3@em0_vlan910: BACKUP -> MASTER (preempting a slower master)
2021-07-07T10:43:59 kernel carp: demoted by -1048576 to 0 (sysctl)
2021-07-07T10:43:59 carp[40586] carp promoted by 1048576 due to service recovery

• Why did CARP status swap in the first place?
• Why does it not swap back until I manually run that code in the shell?

[mimugmail]

You demotion factor was over a million, the reason for it would be of interest.

[clarknova]

I don't see that setting in the CARP config page (attached). I have FRR installed but disabled, so maybe it came from there?

[franco]

Do you have OSFP CARP support enabled in FRR?


Cheers,
Franco

[clarknova]

Do you have OSFP CARP support enabled in FRR?

Yes. Does that have any effect with FRR disabled? Maybe FRR demoted CARP before I disabled FRR?

[mimugmail]

Please disable it and first get your CARP setup stable before you add dynamic routing

[clarknova]

Is it necessary to disable all FRR options when FRR is disabled? Seems odd to have to wipe the whole FRR config rather than just disable FRR.