Amanda-Users

Re: gnutar dies on RedHat 6.2

2002-09-06 20:08:34
Subject: Re: gnutar dies on RedHat 6.2
From: Per olof Ljungmark <peo AT intersonic DOT se>
To: Jon LaBadie <jon AT jgcomp DOT com>
Date: Sat, 07 Sep 2002 01:51:18 +0200
Jon LaBadie wrote:
On Fri, Sep 06, 2002 at 10:55:35AM +0200, Per olof Ljungmark wrote:

At 07:11 9/4/2002 -0400, Gene Heskett wrote:

On Wednesday 04 September 2002 01:52, Per olof Ljungmark wrote:

Hi all,

Got a working amanda 2.4.3b2 installation running on a FreeBSD 4.6
server. AFAICS everything is OK except for one thing, backing up
RedHat 6.2 systems with gnutar. The only trace of the failure I
have is the following in /tmp/amanda/:

sendsize: asking killpgrp to terminate
sendsize: calculating for amname '/etc', dirname '/etc'
sendsize: getting size via gnutar for /etc level 0

and then nothing. Dump is fine from same machine.

Grateful for a hint,
Per olof

Upgrade your 6.2 boxes gnutar to at least 1.13-19?
1.13-25 is the latest, on alpha.gnu.org.

Thanks everybody for your tips. However, I still get the error (or absence of error) above after upgrading to 1.13.25. Is there a way to increase the verbosity of the debug files to find out what is going on? Or can I use sendsize manually on the client? I'm stuck at the moment.


If you look at the runtar debugging files you should see the exact command
lines for tar executed by sendsize.  Note, the tar commands used for size
calculations have their output going to "/dev/null".  The nearly identical
ones used to do the actual backups would be going to stdout, and thus the
filename for output would be "-" (just a dash).

I think they would have to be executed as root on the client.

There are /tmp/amanda debug files on both the client and server.
Have you checked through both for clues?

Just for info, are you also running version 2.4.3b2 amanda on the client system?


Gentlemen,

On the client there are no runtar debugging files, tar is never started AFAICS. On the server, it is just content that those disks are not available and goes on.

For the timeout, this is small non-root partitions on RH 6.2 systems on a diet with very fast disk subsystems. We should be talking a second or two maximum?

Hmmm..., version is 2.4.3b3 on the client...

One thing that could possibly be the cause is the fact that I have tried to persuade the RH systems to have operator as the amanda user as we are mainly on BSD and left something out in doing this. Then the question would be the difference in permissions requirements between dump that works and tar that does not.

Unfortunately I'll be travelling for a few days now so this matter have to rest for a few days... thank you for all your input.

Per olof


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