Amanda-Users

Re: problems with failing (timeout) dupms after upgrade

2003-02-02 06:35:51
Subject: Re: problems with failing (timeout) dupms after upgrade
From: stan <stanb AT awod DOT com>
To: amanda users list <amanda-users AT amanda DOT org>
Date: Sun, 2 Feb 2003 05:49:52 -0500
On Sat, Feb 01, 2003 at 11:04:23AM -0600, Frank Smith wrote:
> --On Saturday, February 01, 2003 08:49:26 -0500 stan <stanb AT awod DOT com> 
> wrote:
> 
> >I'm in the process of changing my Amanda tape server from an HP 9000/735 to
> >an Athaolon 1.2GHZ. I'm using the same HP branded DLT40, and i've upped the
> >holding disk from 2G to 40G. The clients are the same. I've also upgraded
> >(almost all) of the clients to 2.4.3N4.
> >
> >I am having major problems with dumps failing because of "timeouts". What
> >appears to be happening is that the clients are losing there concretion to 
> >the
> >host in mid dump. I;ve decreased the number of dumpers, and put in network
> >bandwidth limiting (using Amanda.conf), and I;ve had a couple of nights
> >where everything went OK. But the whole situation seems awfully "fragile".
> >Before last nights run, i bumped dumpers back up to 16 from 8, and I had 3
> >filessystems fail! Some filessystems on the same machines succeed.
> >
> >I don;t believe i have any specific network issues, as no other network
> >tasks seem to be having problems. The new tapesever is running FreeBSD 4.7
> >STABLE.
> 
> Are you sure it's not a networking related?  No errors from ifconfig
> or on your switch?  Also, check your etimeout and dtimeout settings
> (estimates or data transfer timeouts), maybe one or both need to
> increased.
>  If you're using the same config file as the previous machine,
> and the machine is the only thing that changed, I'd bet on network
> problems (a duplex-mismatch the most likely cause).
> 

The main difference is thta the new machine is a coupe of orders of
magnitude faster. The network is a simple 10base2 cable, and ping -f,
netstat, et all report it as being fine. etimeout is 30000, and drimeoute
is 100000.

What are these numbers stired as? In other words, wjat are the maximum
values?


-- 
"They that would give up essential liberty for temporary safety deserve
neither liberty nor safety."
                                                -- Benjamin Franklin

<Prev in Thread] Current Thread [Next in Thread>