Amanda-Users

Re: Amdump failures in shared cluster

2004-04-20 09:28:31
Subject: Re: Amdump failures in shared cluster
From: Joshua Baker-LePain <jlb17 AT duke DOT edu>
To: Jeff Mulliken <jsmullikenca AT hotmail DOT com>
Date: Tue, 20 Apr 2004 09:23:49 -0400 (EDT)
I moved this discussion to amanda-users, since that's really where it 
belongs.

On Mon, 19 Apr 2004 at 5:37pm, Jeff Mulliken wrote

> 'disklist' file, and the '.amandahosts' file.  The problem, persists, 

By "the problem persists", I'm assuming that you're still getting 'request 
to $FOO timed out'?  Does this setup pass amcheck?

> So, can someone tell me how I get rid of the 'localhost' in the 
> DEFAULT_TAPE_SERVER line?

That's not a big deal.  All that means is that you didn't specify a 
default tape server when you did "./configure".  It comes into play, e.g., 
when you run amrecover, and you can override the default with a command 
line switch (explained in the amrecover man page).

> Secondarily, I suspect that amanda may be colliding with some other software 
> that I have running on this server, that makes extensive use of the internal 
> network.  Is there a way to affect the ports that amanda uses?

Look for the tcpportrange and udpportrange (or some such) options to 
./configure (run './configure --help' to see all the options).

But I'm not convinced that's the problem.  As long as UDP 10080 is open 
(that's where incoming requests go), amanda should pick open ports to use 
for the data transfer.  Try this:

tcpdump -i lo
(In another terminal:)
amcheck -c $CONFIG

And show us the output of the tcpdump.  I assumed in that example that 
your server and client are the same machine.  If they're not, you'll need 
to sniff on eth0 on the server and filter out all the other traffic.

-- 
Joshua Baker-LePain
Department of Biomedical Engineering
Duke University

<Prev in Thread] Current Thread [Next in Thread>
  • Re: Amdump failures in shared cluster, Joshua Baker-LePain <=