Problem with being able to ping outside WAN

[pmcmicha]

I just wanted to bounce this off everyone here. I have a couple of servers that for some reason or another drop communications to the router that they are supposed to go to. At boot up, they communicate just fine, but then after about 5 - 10 minutes, we can no longer ping the router. Now if I delete the arp table entry and then attempt to ping, I can see the router for another 5 - 10 minutes. The dhcp server process is running and all the startup scripts are intact. Does anyone have any idea on how this could be happening? Thanks in advance.

 

[pcunix]

Check your routing (netstat -rn). Betcha $5.00 that something is sending you RIP packets that contain bogus routes.

To cure it, kill the routed process and comment out the routed startup from /etc/tcp

(note that is NOT done by removing routed from the line that sets ALLPROCS- it's the line that acually runs routed)

(also note that routed is NOT what routes packets as more thabn a few people have thought- it is a process that can change routing, but it doesn't DO routing).
.


Tony Lawrence
SCO Unix/Linux Resources

http://pcunix.com

tony@pcunix.com

 

[pmcmicha]

I checked netstat -rn and found that all interfaces show up correct in this table. There are not any extra or misc. entries. The /etc/host, /etc/LAY, and /usr/sbin/ifconfig -a are all correct as well. I've looked throught the osmlog and syslog, but cannot find anything out of the ordinary.

 

[pcunix]

OK, since you saying flushing arp fixes it, then perhaps it is somehow picking up wrong arps. Compare arp -an when it works to arp -an after it doesn't. If the router ip is changing, you may have a dup ip on your network- though that should be showing up in syslog too.

Try forcing it to be permanet with arp -s


Tony Lawrence
SCO Unix/Linux Resources

http://pcunix.com

tony@pcunix.com