Picon
Favicon

Re: mondorescue: no space left on device

Hi Alan,

In all the start_udev script versions (RHEL 5, RHEL 5 modified, and RHEL 6), LANG PTSDIR and SHMDIR
variables are not defined before the awk section (line #136) ; so if you put the echoes before that line it's
normal to see these variables empty.

LANG is defined just in the beginning of the awk line.

The "mktemp -d" command creates a directory and returns the directory name.
# mktemp -d
/tmp/tmp.h7g2EEOjgs
# ls -ld /tmp/tmp.h7g2EEOjgs 
drwx------. 2 root root 4096 Apr  2 11:01 /tmp/tmp.h7g2EEOjgs

In the awk section, PTSDIR and SHMDIR are defined in all the start_udev script versions (RHEL 5, RHEL 5
modified, and RHEL 6) through:
PTSDIR=$(mktemp -d)    (line #138)
SHMDIR=$(mktemp -d)    (line #142)

By putting your echoes just after these lines you should see these variables contents.
After doing that, if your echoes still show PTSDIR and SHMDIR variables empty, it could be because in your
case /tmp is full and then mktemp doesn't succeed and returns an empty string.

Rgds,
Victor

-----Original Message-----
From: Alan Pittman [mailto:Alan.Pittman <at> publix.com] 
Sent: vendredi 30 mars 2012 22:44
To: 'Mondo mailing list'
(Continue reading)

Jeffrey D. Fulmer | 2 Apr 2012 16:04
Favicon

Re: mondorescue: no space left on device

I'm having the same problem on RHEL 5.7. I'll try the script and report back.

________________________________________
From: Gattegno, Victor (CPR&Q GCC - ISS/SW Tower) [victor_gattegno <at> hp.com]
Sent: Thursday, March 29, 2012 10:03 AM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mondorescue: no space left on device

Nice try Alan, yes it's strange that you didn't see any of your echoes...

Interesting Stefan, I added that information (no more problem when mindi downgraded to 2.1.0) in mondo
Ticket #600 that I created for that problem.
http://trac.mondorescue.org/ticket/600

I checked mindi 2.1.0 rcS shell-script file, it differs little from the one of mindi 2.1.1, and both call the
RHEL start_udev script...

I saw also that ChangeLog file of mindi 2.1.0 is empty. Hopefully, in NEWS file we can see the changes between
mindi 2.1.1 and 2.1.0.

>From my side, I modified the tmpfs mount section of the RHEL5 start_udev script. I attach the modified
start_udev to this email, if some of you want to try it on a RHEL 5 machine with mindi 2.1.1.

Rgds.

------------------------------------------------------------------------------
This SF email is sponsosred by:
Try Windows Azure free for 90 days Click Here 
http://p.sf.net/sfu/sfd2d-msazure
(Continue reading)

Jeffrey D. Fulmer | 2 Apr 2012 20:29
Favicon

Re: mondorescue: no space left on device

I placed the fixed start_udev script in /sbin and built a new mondorescue.iso.  This had no effect. I still
got no space left on device.

________________________________________
From: Jeffrey D. Fulmer [JDFulmer <at> Armstrong.com]
Sent: Monday, April 02, 2012 10:04 AM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mondorescue: no space left on device

I'm having the same problem on RHEL 5.7. I'll try the script and report back.

________________________________________
From: Gattegno, Victor (CPR&Q GCC - ISS/SW Tower) [victor_gattegno <at> hp.com]
Sent: Thursday, March 29, 2012 10:03 AM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mondorescue: no space left on device

Nice try Alan, yes it's strange that you didn't see any of your echoes...

Interesting Stefan, I added that information (no more problem when mindi downgraded to 2.1.0) in mondo
Ticket #600 that I created for that problem.
http://trac.mondorescue.org/ticket/600

I checked mindi 2.1.0 rcS shell-script file, it differs little from the one of mindi 2.1.1, and both call the
RHEL start_udev script...

I saw also that ChangeLog file of mindi 2.1.0 is empty. Hopefully, in NEWS file we can see the changes between
mindi 2.1.1 and 2.1.0.

>From my side, I modified the tmpfs mount section of the RHEL5 start_udev script. I attach the modified
(Continue reading)

Heijmans S (spir-it | 3 Apr 2012 13:11
Picon

buffer rpm not signed

Hello,

Would it be possible to also sign the following buffer rpm's;
ftp://ftp.mondorescue.org/rhel/5/x86_64/buffer-1.19-4.rhel5.x86_64.rpm
ftp://ftp.mondorescue.org/rhel/6/x86_64/buffer-1.19-4.rhel6.x86_64.rpm

# rpm -qpi buffer-1.19-4.rhel6.x86_64.rpm
Name        : buffer                       Relocations: (not relocatable)
Version     : 1.19                              Vendor: (none)
Release     : 4.rhel6                       Build Date: Fri 18 Feb 2011 12:18:28 AM CET
Install Date: (not installed)               Build Host: localhost
Group       : Applications/Archiving        Source RPM: buffer-1.19-4.rhel6.src.rpm
Size        : 45710                            License: GPL
Signature   : (none)
Packager    : Bruno Cornec <bruno <at> project-builder.org>
URL         : http://hello-penguin.com/software/buffer
Summary     : This program speeds up writing tapes on remote tape drives
Description :
This is a program designed to speed up writing tapes on remote tape drives.
When this program is put "in the pipe", two processes are started.
One process reads from standard-in and the other writes to standard-out.
Both processes communicate via shared memory.
# rpm -qpi buffer-1.19-4.rhel5.x86_64.rpm
Name        : buffer                       Relocations: (not relocatable)
Version     : 1.19                              Vendor: (none)
Release     : 4.rhel5                       Build Date: Fri 18 Feb 2011 12:18:31 AM CET
Install Date: (not installed)               Build Host: localhost.localdomain
Group       : Applications/Archiving        Source RPM: buffer-1.19-4.rhel5.src.rpm
Size        : 39338                            License: GPL
Signature   : (none)
(Continue reading)

Bruno Cornec | 3 Apr 2012 15:34
Picon
Favicon

Re: buffer rpm not signed

Heijmans S (spir-it) said on Tue, Apr 03, 2012 at 01:11:52PM +0200:

> Would it be possible to also sign the following buffer rpm's;
> ftp://ftp.mondorescue.org/rhel/5/x86_64/buffer-1.19-4.rhel5.x86_64.rpm
> ftp://ftp.mondorescue.org/rhel/6/x86_64/buffer-1.19-4.rhel6.x86_64.rpm

Done.

Bruno.
--

-- 
Open Source & Linux Profession Lead EMEA         http://opensource.hp.com
EMEA Open Source HPISC Chief Technologist        http://hpintelco.net
FLOSS projects:     http://mondorescue.org     http://project-builder.org 
Musique ancienne? http://www.musique-ancienne.org http://www.medieval.org

------------------------------------------------------------------------------
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second 
resolution app monitoring today. Free.
http://p.sf.net/sfu/Boundary-dev2dev
Heijmans S (spir-it | 3 Apr 2012 16:19
Picon

Re: buffer rpm not signed

Hi Bruno,

Thanks.
Just noticed that afio & mindi-busybox also are not signed, could you sign those as well?

Stefan

-----Oorspronkelijk bericht-----
Van: Bruno Cornec [mailto:Bruno.Cornec <at> hp.com]
Verzonden: dinsdag 3 april 2012 15:34
Aan: Mondo mailing list
Onderwerp: Re: [Mondo-devel] buffer rpm not signed

Heijmans S (spir-it) said on Tue, Apr 03, 2012 at 01:11:52PM +0200:

> Would it be possible to also sign the following buffer rpm's;
> ftp://ftp.mondorescue.org/rhel/5/x86_64/buffer-1.19-4.rhel5.x86_64.rpm
> ftp://ftp.mondorescue.org/rhel/6/x86_64/buffer-1.19-4.rhel6.x86_64.rpm

Done.

Bruno.
--
Open Source & Linux Profession Lead EMEA         http://opensource.hp.com
EMEA Open Source HPISC Chief Technologist        http://hpintelco.net
FLOSS projects:     http://mondorescue.org     http://project-builder.org
Musique ancienne? http://www.musique-ancienne.org http://www.medieval.org

Informatie van de Raad voor de rechtspraak, de rechtbanken, de gerechtshoven en de bijzondere colleges
vindt u op www.rechtspraak.nl.
(Continue reading)

Alan Pittman | 3 Apr 2012 16:31
Favicon

Re: mondorescue: no space left on device

Hi Victor,
   Thanks for the update. I've modified my script accordingly and added a few more echoes in the
make_extra_nodes function. See the listings below. Look for the "<<--" symbols.

After running archive and burning a new image, this is the output I get during a boot of the DVD:

 udev_root = /dev
 read alanx

 in make_extra_nodes
 sh: /dev/MAKEDEV: unknown operand
 /sbin/start_udev: Line 201: -|: not found
 /sbin/start_udev: Line 201: -|: not found
 /sbin/start_udev: Line 201: -|: not found
 /sbin/start_udev: Line 201: -|: not found
 /sbin/start_udev: Line 201: -|: not found
 /sbin/start_udev: Line 201: -|: not found
 Error: /etc/makedev.d is not a directory
 Error: /etc/makedev.d is not a directory
 ls: /etc/udev/device: No such file or directory

  Filesystem     1K-Blocks     Used   Available    Used%    Mounted-on
     None             5120       0       5120        0        /dev
     None          1029292       0    1029292        0        /dev/shm

  Followed by many: no space left on device 

When I boot RHEL5.8 natively, I only get the following:

udev_root = /dev
(Continue reading)

Jeffrey D. Fulmer | 3 Apr 2012 20:30
Favicon

Re: mondorescue: no space left on device

I swapped in a start_udev from a RHEL 6 box on the RHEL 5.7 box I'm developing my recovery methods on. With this
file, I was able to recover the server. It wouldn't be a big deal to distribute that on servers that require
bare metal restoration but I wonder if it wouldn't be better to pull one from 5.2 (I know that works with
mondo). 

I did have an issue after recovery: I'm 99% certain the system I archived had SELINUX=permissive set in
/etc/selinux/config and when I restored it, it was set to enabled. After the server came back up, I was
unable to login until I booted with selinux=0  Does that sound right? We use system designer and every
server is set to permissive the moment it comes up. I have a high degree of certainty that it was permissive
on this system.

Any thoughts?

________________________________________
From: Jeffrey D. Fulmer [JDFulmer <at> Armstrong.com]
Sent: Monday, April 02, 2012 2:29 PM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mondorescue: no space left on device

I placed the fixed start_udev script in /sbin and built a new mondorescue.iso.  This had no effect. I still
got no space left on device.

________________________________________
From: Jeffrey D. Fulmer [JDFulmer <at> Armstrong.com]
Sent: Monday, April 02, 2012 10:04 AM
To: Mondo mailing list
Subject: Re: [Mondo-devel] mondorescue: no space left on device

I'm having the same problem on RHEL 5.7. I'll try the script and report back.

(Continue reading)

Heijmans S (spir-it | 3 Apr 2012 21:28
Picon

Re: mondorescue: no space left on device

Hi Jeffrey,

>>After the server came back up, I was unable to login until I booted with selinux=0
Make sure you perform an SELinux autorelabel after the mondoerestore has finished, we do it like this on RHEL5;

# mount /dev/VolGroup00/LogVol00 /mnt
# cd mnt
# touch .autorelabel
# cd /
# umount /mnt
# exit

Stefan

Informatie van de Raad voor de rechtspraak, de rechtbanken, de gerechtshoven en de bijzondere colleges
vindt u op www.rechtspraak.nl.

------------------------------------------------------------------------------
Better than sec? Nothing is better than sec when it comes to
monitoring Big Data applications. Try Boundary one-second 
resolution app monitoring today. Free.
http://p.sf.net/sfu/Boundary-dev2dev
pclark | 3 Apr 2012 21:12

Re: mondorescue: no space left on device

I've had to put an .autorelabel file in / before, but if it's permissive or off it shouldn't have to be done. 

What's the content of /sys/fs/selinux/enforce on the box you backed up?

On Apr 3, 2012, at 2:30 PM, "Jeffrey D. Fulmer" <JDFulmer <at> Armstrong.com> wrote:

> I swapped in a start_udev from a RHEL 6 box on the RHEL 5.7 box I'm developing my recovery methods on. With
this file, I was able to recover the server. It wouldn't be a big deal to distribute that on servers that
require bare metal restoration but I wonder if it wouldn't be better to pull one from 5.2 (I know that works
with mondo). 
> 
> I did have an issue after recovery: I'm 99% certain the system I archived had SELINUX=permissive set in
/etc/selinux/config and when I restored it, it was set to enabled. After the server came back up, I was
unable to login until I booted with selinux=0 Does that sound right? We use system designer and every
server is set to permissive the moment it comes up. I have a high degree of certainty that it was permissive
on this system.
> 
> Any thoughts?
> 
> 
> ________________________________________
> From: Jeffrey D. Fulmer [JDFulmer <at> Armstrong.com]
> Sent: Monday, April 02, 2012 2:29 PM
> To: Mondo mailing list
> Subject: Re: [Mondo-devel] mondorescue: no space left on device
> 
> I placed the fixed start_udev script in /sbin and built a new mondorescue.iso. This had no effect. I still
got no space left on device.
> 
> 
(Continue reading)


Gmane