Greets, amanda-users,
I use amanda-3.1.2 for a few days now, on my servers and at a customer.
Now I see problems there. Or at least questions ...
The cron-job started there results in a failed amdump-run, all DLEs show
"FAILED [data write: Broken pipe]" in the report email.
I then logged into
On Thu, Aug 19, 2010 at 5:30 AM, Stefan G. Weichinger wrote:
> Wed Aug 18 21:09:01 2010: sendbackup: critical (fatal): index tee cannot
> write [Broken pipe]
...
> We use a wrapper there for gnu-tar ..
That means that Amanda got an error writing to the input of
"/usr/local/bin/amgtar -tf - 2>/d
Am 19.08.2010 16:37, schrieb Dustin J. Mitchell:
> On Thu, Aug 19, 2010 at 5:30 AM, Stefan G. Weichinger wrote:
>> We use a wrapper there for gnu-tar ..
>
> That means that Amanda got an error writing to the input of
> "/usr/local/bin/amgtar -tf - 2>/dev/null | sed -e 's/^\.//'"
> which is used
On Thu, Aug 19, 2010 at 10:32 AM, Stefan G. Weichinger wrote:
> And why does it work then at the second time, when I start amdump
> manually? The amount of input should be the same at the 2nd time ... ?
That's why I suspect it's a race condition. Recall we just found out
about the GNU tar race c
Am 19.08.2010 18:05, schrieb Dustin J. Mitchell:
> On Thu, Aug 19, 2010 at 10:32 AM, Stefan G. Weichinger
> wrote:
>> And why does it work then at the second time, when I start amdump
>> manually? The amount of input should be the same at the 2nd time ... ?
>
> That's why I suspect it's a race c
Am 19.08.2010 19:01, schrieb Dustin J. Mitchell:
> On Thu, Aug 19, 2010 at 11:45 AM, Stefan G. Weichinger
> wrote:
>> What could I do to spot the problem?
>> Show the wrapper? Browse for other logs?
>
> Well, first, reproduce it - either determine that it's intermittent,
> or that it only occurs
On Thu, Aug 19, 2010 at 11:45 AM, Stefan G. Weichinger wrote:
> What could I do to spot the problem?
> Show the wrapper? Browse for other logs?
Well, first, reproduce it - either determine that it's intermittent,
or that it only occurs when run from cron. Narrow down the cause as
much as possibl
On Thu, Aug 19, 2010 at 12:26 PM, Stefan G. Weichinger wrote:
> We'll see what the cronjob will do ... What might be the difference?
Environment variables, usually. Some systems have other context
(e.g,. Mach-based systems) that is inherited with a process. Selinux
might be involved, but I assu
Am 19.08.2010 19:45, schrieb Dustin J. Mitchell:
> On Thu, Aug 19, 2010 at 12:26 PM, Stefan G. Weichinger
> wrote:
>> We'll see what the cronjob will do ... What might be the difference?
>
> Environment variables, usually. Some systems have other context
> (e.g,. Mach-based systems) that is inh
On Fri, Aug 20, 2010 at 04:22:56PM +0200, Stefan G. Weichinger wrote:
> Am 19.08.2010 19:45, schrieb Dustin J. Mitchell:
> > On Thu, Aug 19, 2010 at 12:26 PM, Stefan G. Weichinger
> > wrote:
> >> We'll see what the cronjob will do ... What might be the difference?
> >
> > Environment variables,
Am 20.08.2010 18:46, schrieb Jon LaBadie:
>> Does anyone recommend adding such a cron-wrapper which sources
>> .profile and .bashrc ?
>
> I'd advise against sourcing .profile and/or any 'rc' files. They
> typically contain items of use in an interactive session, think
> settings for PS[1-4], alia
Am 20.08.2010 18:58, schrieb Stefan G. Weichinger:
> I would only like to somehow compare the environment variables, just to
> narrow down the reasons ...
>
> Would crontabbing something like "env >> my_result" work?
Tried that. Big difference.
examples:
result_crontab:SHELL=/bin/sh
result_she
On Fri, Aug 20, 2010 at 1:13 PM, Stefan G. Weichinger wrote:
> Tried that. Big difference.
So try running amdump by hand with the crontab environment.
Dustin
--
Open Source Storage Engineer
http://www.zmanda.com
Am 20.08.2010 20:27, schrieb Dustin J. Mitchell:
> On Fri, Aug 20, 2010 at 1:13 PM, Stefan G. Weichinger
> wrote:
>> Tried that. Big difference.
>
> So try running amdump by hand with the crontab environment.
I did and in consequence I also edited the cronjob for amanda to set a
whole bunch of v
Am 23.08.2010 22:00, schrieb Stefan G. Weichinger:
> I did and in consequence I also edited the cronjob for amanda to set a
> whole bunch of variables reported by env within amanda's shell.
detail question:
while searching around I somewhere read that the amanda-user should have
the shell /bin/s
On Mon, 23 Aug 2010 22:30:02 +0200
"Stefan G. Weichinger" wrote:
> Dustin, could you let me know which shell(s) are OK for the amanda
> user?
>
> bash seems more comfortable somehow, I have PATH and PS1 etc already
> set up OK.
You should be OK with bash.
In fact on Centos 5.5, sh is a symlink
Am 23.08.2010 22:49, schrieb Charles Curley:
> You should be OK with bash.
>
> In fact on Centos 5.5, sh is a symlink to bash:
>
> # ll $(which sh)
> lrwxrwxrwx 1 root root 4 Jun 7 21:40 /bin/sh -> bash
>
> On Ubuntu 8.04 (Ubuntu 8.04.4 LTS, to be pedantic), sh is a symlink to
> dash. I changed
On Mon, Aug 23, 2010 at 02:49:18PM -0600, Charles Curley wrote:
> On Mon, 23 Aug 2010 22:30:02 +0200
> "Stefan G. Weichinger" wrote:
>
> > Dustin, could you let me know which shell(s) are OK for the amanda
> > user?
> >
> > bash seems more comfortable somehow, I have PATH and PS1 etc already
> >
Am 24.08.2010 06:07, schrieb Jon LaBadie:
> On Mon, Aug 23, 2010 at 10:00:43PM +0200, Stefan G. Weichinger wrote:
> [snip]
>>
>> Part of nr. 2: post the list and listen for feedback
>>
>> See yourself:
>>
> I'll take a shot and comment.
Quick reply without detail:
Thanks, Jon, for your comments!
Am 24.08.2010 22:31, schrieb Stefan G. Weichinger:
> amdump-cronjob runs fine right now, I will tune more tomorrow.
Another short report:
The currently working crontab contains these variables:
HOSTNAME=juno
HOST=juno
USER=amanda
XNLSPATH=/usr/X11R6/lib/X11/nls
HOSTTYPE=x86_64
PATH=/usr/sbin:/us
20 matches
Mail list logo