On Mon, May 21, 2012 at 11:19:11AM +0000, Matt Hamilton wrote: > Hi All, > I've recently setup up a series of 6 OpenBSD boxes all running 5.1/amd64 > and connected together via an HP switch. The all run ospfd and bgpd. > They each connect out to different external networks and most speak BGP > to external peers. > > I keep seeing bgpd just quitting of its own accord. In the logs I see > something like this: > > > May 20 12:28:42 janet-fw1 bgpd[18874]: dispatch_rtmsg[change] mpath > route not found > May 20 12:28:42 janet-fw1 bgpd[32738]: neighbor 213.133.64.98 (Core > Router 2): sending notification: Cease, administratively down > May 20 12:28:42 janet-fw1 bgpd[32738]: neighbor 213.133.64.97 (Core > Router 1): sending notification: Cease, administratively down > May 20 12:28:42 janet-fw1 bgpd[32738]: neighbor 195.10.248.38 (Bristol > City Council): sending notification: Cease, administratively down > May 20 12:28:42 janet-fw1 bgpd[32738]: neighbor 195.10.248.34 (SWERN): > sending notification: Cease, administratively down > > I don't know what is going on, but there is always that mpath error there > before it cuts out. I don't seem to be able to spot a pattern in the > timing, it just cuts out. > > Any ideas? Anyone else seen this? >
The "dispatch_rtmsg[change] mpath route not found" is a fatal error (bgpd quits because of this). The problem seems to be a multipath route that is changed but bgpd can not find the route in its own table and freaks out. I have not seen this happen so it seems to depend on the setup. Currently I'm unsure how that can happen, it looks like bgpd missed some routing updates. -- :wq Claudio