Amanda-Users

Re: Problems with 4.5 and Amanda 2.6.1

2009-08-20 16:30:10
Subject: Re: Problems with 4.5 and Amanda 2.6.1
From: stan <stanb AT panix DOT com>
To: Chris Hoogendyk <hoogendyk AT bio.umass DOT edu>
Date: Thu, 20 Aug 2009 16:23:10 -0400
On Thu, Aug 20, 2009 at 12:54:06PM -0400, Chris Hoogendyk wrote:
> 
> 
> stan wrote:
> >I recently tried to upgrade several of my OpenBSD machines to 4.5 (X86). I
> >also upgraded the Amanda clients on them to 2.6.1, which I have working on
> >FreeBSD, Linux, Solaris 5, Solaris 8, and HP-UX. 
> >
> >Despite my best efforts, I have yet to get the OpenBSD 4.5 clients to back
> >up correctly. All the configurations are set up to match what has been done
> >on FreeBSD. Amcheck works fine, and when I start a backup, the size
> >estimates step works. But 80% of the time the actual dumps fail. They
> >start, but fail in mid dump. They are returning a EAGAIN error on the
> >network pipes back to the Amanda master machine. I have connected one of 
> >the
> >OpenBSD machines directly to the Amanda master with a crossover cable to
> >eliminate any possible firewall/router et all issues.
> >
> >I notice that the port of Amanda is quite an old version, and the
> >developers of this project are (unfortunately) breaking compatibility with
> >older clients. I don't like this, but, at this point in time, it is beyond
> >my resources to address this breakage.
> >
> >Has anyone gotten an newer version of Amanda to work on OpenBSD 4.5? Can
> >anyone suggest what I should do to try to diagnose this problem further?
> 
> Sounds like you've had a fair bit of experience on various platforms 
> implementing Amanda.
> 
> Since you upgraded the OpenBSD, perhaps for sanity and just checking and 
> troubleshooting, you should grab an older Amanda that you already have 
> implemented, and get that working on the upgraded OpenBSD. That might 
> just work, or it might clue you in to what's going on.

One more thing thta's going to make this df`ificult is that the contractor
I had do this upgrade installed 86 OpenBSD, when we had AMD64 previously.
So my old binaries are not any good either :-(

-- 
One of the main causes of the fall of the roman empire was that, lacking
zero, they had no way to indicate successful termination of their C
programs.