My previous company has a pair of firewalls running OpenBSD 4.4 with CARP. They've been running with no problem since just after the 4.4 release, until the last couple of days.
Now, the firewall that should be in BACKUP state has somehow decided that it needs to be MASTER for some, but not all, of the CARP interfaces, even though the master machine is running fine. Something like this: if machine 1 machine 2 carp0 MASTER BACKUP carp1 MASTER BACKUP carp2 MASTER MASTER carp3 MASTER BACKUP carp4 MASTER MASTER The interfaces where both machines try to be MASTER at the same time become unreliable or unreachable. I looked around Google but couldn't turn up any reports of similar issues. Admittedly I might have been searching for the wrong terms, though. Any ideas as to what could be causing this problem? They're likely to rebuild both machines in the next week or so, either with 4.6 (so they can keep their existing pf.conf) or with 4.9 so as to be current, but they'd like some assurance that a rebuild will actually solve the problem. (If it were, say, a failing NIC, updating the software wouldn't help.) For whatever it's worth, the machines in question are Poweredge R200s, with the two on-board Broadcom gigabit ports and an additional Intel gigabit card for pfsync. They're running the i386 rather than the amd64 version of OpenBSD. Thanks in advance for any suggestions.