Server is initializing itself

Keith Woodworth kwoody at citytel.net
Fri Feb 10 18:39:46 UTC 2006


On Fri, 10 Feb 2006, Aleksander wrote:

|->Keith Woodworth wrote:
|->
|->>damn everything is the same, IP, GW, mask, I can ping the default gateway
|->>for that vlan, but nothing is going past that GW to the next hop into my
|->>upstream.
|->>
|->>
|->Is the default route (gateway) set on the FBSD box? Maybe the GW has a
|->limitation on whom to SNAT based on the MAC address which I doubt,
|->assuming the GW separates your LAN from the Internet.

I'm not sure what happened but I did get it working about 15 mins after
this post.

I went to the console of the machine, netstat -rn, there was no default
route set, even though rc.conf has one in there and I thought I'd checked
it when first starting to trouble shoot this since I could ping local
machines, but nothing past the GW.

As a test, I put the machine on the same vlan with a different IP on a
different port so the GW would be the same. Manually added a default route
and BIND came up right away.

Changed the IP on the machine back to what it was in the first place, did
a netstat -rn and the default route went away, even though the IP's are
on the same vlan/subnet. I'm not sure if Ive seen this behaviour before.
Maybe Ive just not paid attention enough. :)

But added the route back in, moved the machine back to the proper switch
port, did a clear arp on the switch and everything worked.

Though when doing some pings at the time I got ICMP redirect messages on
the 2nd or 3rd reply from the switch, then things went back to normal.

The upstream provider does some NAT internally, as shown by this trace:

 2  router (204.244.99.65)  2.196 ms  2.007 ms  4.104 ms
 3  10.204.23.13 (10.204.23.13)  14.052 ms  8.036 ms  12.761 ms
 4  10.204.2.109 (10.204.2.109)  29.527 ms  32.804 ms  22.895 ms
 5  ge-0-0-0-2.vsb1.navigata.net (204.244.12.13)  29.474 ms  25.002 ms

And it did occur to me that maybe their stuff was doing something funky
but I couldnt think of where or why it might. I think though it had to
simply to do with arp caches, default gateways and the confusion that the
box was accessable on the local network via SSH, but could not get out.

Thanks,
Keith





More information about the bind-users mailing list