I have two CARPed boxes that connect to upstream providers, and do BGP routing with them. CARP is configured in active/passive setup. Network between me and each provider is /29. Setup is described in detail here (although in Serbian, but schema in sidebar and conf files are universal):
https://www.mimar.rs/openbsd-na-obodu-korporacijske-mreze/ I have BGP neighbors setup by help of local-address and demote carp, and in case of manual restart of master firewall i have graceful failover (internet radio stream does not even hiccup). After master reboots, it waits for bgp session to settle (demote carp), and after that it takes function of master. No hiccups here either. However, sometimes failover happens without any apparent reason, with both my boxes running. In that case, master never re-claims its master status automatically. Any comments? How can I see in logs why master became backup? And why doesn't it re-claim its master status even though it has BGP session active? -- Marko Cupać
