Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 09:41:35 -0400
We are currently looking for ways to take mtx out of the mix with Amanda. The problem has been that we don't have anyone on the Amanda team that understands SCSI and kernel SCSI interfaces well enoug
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 09:35:18 -0400
Not to be coy, but I would guess that it's the significant changes you're making to your network! If I recall, you're using UDP-based auth, so perhaps these network changes are causing packet loss an
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 09:37:50 -0400
You're using in-memory splitting with a 1G splitsize, which is causing the memory allocation. Check your splitting params (split_diskbuffer, taper_splitsize, and fallback_splitsize). See the latest a
That's sort of the point. I need to be able to docment the brain damage that is being done to the network :-) WE are largely BSDTCP auth at this poit, as UPD stoped working a couple of months agao. -
Author: Charles Curley <charlescurley AT charlescurley DOT com>
Date: Thu, 24 Sep 2009 09:22:06 -0600
Thanks. Since this is running on a fit-PC 1.0 (http://www.fit-pc.com/fit-pc1/fit-pc-1-0-specifications.html), with 229 MB of physical memory and 760 MB of swap, I reduced the two splitsize values, fr
Right, you and I are thinking alike on this. What I was hoping was that there was the functional equivilant of a -v (verbose) flag to amcheck, so that I could see what it was doing. I'd prefer not to
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 09:06:00 -0400
Can you attach the trace log so we can see what happened after that? And also the taper's debug log. Dustin -- Open Source Storage Engineer http://www.zmanda.com
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 12:24:08 -0400
Well, there's no need to allocate space for both a memory and disk buffer -- the memory buffer is only used if the disk buffer is incorrectly configured. Dustin -- Open Source Storage Engineer http:/
FYI, an amcheck of 53 machines (all of which pass) take a reliable 6 minutes. Historically, depending on whether we had read the tape before or not, the server could take up to 3 seconds, or less tha
Author: Steven Backus <backus AT whimsy.med.utah DOT edu>
Date: Thu, 24 Sep 2009 08:38:38 -0600 (MDT)
And I say: I'm using gnutar and my performance has gone to hell since the 5.4 version of RHEL came out. What do we do now? Steve -- Steven J. Backus Computer Specialist University of Utah E-Mail: st
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Thu, 24 Sep 2009 11:32:17 -0400
If it's still TCP then the best thoughts I have are to look at ARP and DNS resolution times. Has a resolver been removed that one of your machines is still using? Dustin -- Open Source Storage Engine
Author: Nathan Stratton Treadway <nathanst AT ontko DOT com>
Date: Thu, 24 Sep 2009 23:10:11 -0400
On my system (Amanda 2.5.1p3), the amcheck log files found at /var/log/amanda/server/<CONFIG>/amcheck.*.debug do include timestamp information for when each remote host sent its reply package. It onl
Author: Gene Heskett <gene.heskett AT verizon DOT net>
Date: Fri, 25 Sep 2009 10:23:35 -0400
This was yesterdays 20090924 snapshot, so this is not fixed yet. Amcheck however is happy as a clam. This last worked on my linux box at 2.6.2alpha- 20090831. Which I'm going to re-install & run a ba
Author: "Dustin J. Mitchell" <dustin AT zmanda DOT com>
Date: Fri, 25 Sep 2009 14:50:24 -0400
On Fri, Sep 25, 2009 at 10:23 AM, Gene Heskett <gene.heskett AT verizon DOT We've all been on vacation for the last week, so again, I'm not surprised it's still broken :) I'll let you know when we (t
Hi Dustin, I hope you have recovered from your shock regarding threads on OpenBSD ;-) Here's an idea for a partial fix; I have no idea how easy or difficult it would be. Until threads on OpenBSD work
Author: Gene Heskett <gene.heskett AT verizon DOT net>
Date: Fri, 25 Sep 2009 16:41:02 -0400
wrote: I hope you worked on your tan, you found a new recipe for whatever your favorite poison is & the chiggers didn't lunch on you excessively. Off topic, way off topic: While you were on vacation,
Thanks, that should help troubeshoot this. -- 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.
Author: Jean-Francois Malouin <malin AT bic.mni.mcgill DOT ca>
Date: Sat, 26 Sep 2009 13:06:35 -0400
* Dustin J. Mitchell <dustin AT zmanda DOT com> [20090908 13:08]: I'm reviving this thread as I got hit again last night on a server running 2.6.1p1 that I reinstalled with REP_TIMEOUT=(12*60*60) ie,
Author: Gene Heskett <gene.heskett AT verizon DOT net>
Date: Sat, 26 Sep 2009 22:26:37 -0400
Greetings all; I am attempting to add another machine to my disklist, and I have installed the amanda-client-2.4.5p1 on that machine, but I can't find where it might be started, no can I find any con
Hi all, I've been happily using Amanda for some time now, but I seem to have run into a recent snag. The backup server has multiple addresses on a single interface and, for some reason, it has taken