Bug 873
| Summary: | Invalid route to remote network | ||
|---|---|---|---|
| Product: | Base System | Reporter: | Andrew Webster <andreww> |
| Component: | bin | Assignee: | Bill Fenner <fenner> |
| Status: | Closed FIXED | ||
| Severity: | Affects Only Me | ||
| Priority: | Normal | ||
| Version: | 2.1-STABLE | ||
| Hardware: | Any | ||
| OS: | Any | ||
Responsible Changed From-To: freebsd-bugs->wollman State Changed From-To: open->feedback "routed" could install completely bogus netmasks, which could potentially cause behavior like this. Could you try the routed from FreeBSD-stable (just get src/usr.sbin/routed/inet.c and rebuild) and see if the symptoms change? Responsible Changed From-To: wollman->fenner fenner recently fixed routed netmask stuff Andrew, Would you mind giving the routed from -stable a try? You can just grab ftp://ftp.freebsd.org/pub/FreeBSD/FreeBSD-stable/src/usr.sbin/routed/inet.c and rebuild your routed. I fixed a problem which almost definitely caused what you were seeing. If you tried this and replied to me already, sorry for having lost your reply & could you send it again? Thanks, Bill State Changed From-To: feedback->closed Closed after 7 months of no feedback. |
On bootup, iijppp is started in -auto mode to initiate ppp connection over tun0 device, default route to 198.168.61.61 is established by iijppp Routed is started in -q mode. A netware fileserver is broadcasting RIP for 198.168.37.0 recahable through 198.168.36.99 and 198.168.38.0 recahable through 198.168.36.99. When I attempt to traceroute to 198.168.54.8 (ns.rezonet.net) a host outside our domain. One of two things will happend: 1. If routed -q is running, I get this result: 55 [vhf:/home/andreww] : traceroute 198.168.54.8 traceroute to 198.168.54.8 (198.168.54.8), 30 hops max, 40 byte packets 1 fs (198.168.36.99) 2.483 ms !H 1.111 ms !H 0.964 ms !H Netstat -rna produces (other stuff removed): default 198.168.61.61 UGc 20 3 tun0 2. If routed is NOT running, I get this result: 52 [vhf:/home/andreww] : traceroute ns.rezonet.net traceroute to ns.rezonet.net (198.168.54.8), 30 hops max, 40 byte packets 1 S496.InterLink.NET (198.168.61.61) 226.432 ms 191.383 ms 198.247 ms 2 MIX.NET (198.168.73.2) 197.366 ms 205.749 ms 168.272 ms 3 ns.RezoNet.NET (198.168.54.8) 197.555 ms 205.701 ms 198.371 ms Netstat -rna produces (other stuff removed): 198.168.54.8 198.168.61.61 UGHW 2 69 tun0 I've put a protocol tracer on both the ppp link and our internal networks and there are no RIP packets with any information that could cause this erratic behaviour. Fileserver and ppp link are clean. Fileserver's RIP packets are okay. Fix: Don't run routed -q, use static routes. Routing information corrupted by routed? How-To-Repeat: As long as routed -q is run on bootup this happends.