> I assume this is a /tmp/amanda/amandad*debug from the failnig 
> client? 

Yes, sorry about that.

 It 
> looks like it's sending the estimates, but not hearing back from the 
> server.  If that's the case, then I think the possiblities are:
> 
> 1) The server is no longer listening for a response, i.e. the 
> client took
>    too long to do the estimates.  What time do you kick off 
> your amdump?  
>    Is it more than 1800 seconds before 2:28 AM?

I kick off my dump at 2:30am. Through all of this I have found that my
servers are running a few minutes apart. The client is 3 minutes behind
the server...so that would mean it kicked off at 2:27 according to the
client. 2:27.04 to be exact...and timed out at 2:28.50. Now I realize I
don't know the 'new math' ... but that still doesn't seem like 1800
seconds. ;) ??? 
> 
> 2) The server can't hear the response from the client becuase 
> of firewall 
>    issues.  You said the server is RH8 -- make double/triple 
> sure that you
>    don't have ipchains or iptables running.  IIRC, RH8 *still* uses 
>    ipchains.  Look at 'chkconfig --list', 'ipchains -nL', and the 
>    contents of /etc/sysconfig/ipchains (adjusting as appropriate for
>    iptables if I'm wrong about that).

In chkconfig --list...iptables was turned on for runlevels 2,3,4,5...but
there were no rules in iptables -L. I killed it just in case. ipchains
is nowhere to be found.
> 
> 3) Any sort of firewalling on the client side -- I know diddly about 
>    Solaris.

The box is as hardened as it can be for a web server, but that's about
it. Again...none of this has changed from two weeks ago.