Getting Mondo to work on a trixbox
Submitted by AndyInNYC on Sat, 12/08/2007 - 12:45pm.
I'm trying to get a clean backup of my box so that I can keep trying to get the zapMicro zma800p installed and working without having to completely reinstall from scratch each time one of their 'tech' suggestions or something I find on the web leads me astray.
My install of Mindi and Mondo (both from the versions suggested in the post on this site and the newest available) do a clean backup but on a 'compare' report during the boot that "error while loading shared libraries: libpthread.so.0. Cannot open shared object file: No such file or directory......"
I found a reference to mondo and this error, but it wasn't in English and I'm an ugly American (and ugly per a few ex-girlfriends). From what I understood of the translation, certain libraries aren't imported into some directory where they need to be - perhaps a Centos fault.
Any suggestions from the experts here? Is there a Mondo forum somewhere?
Andrew
Mondo backup / restore
Sat, 12/08/2007 - 1:27pm
A while ago I wrote a Mondo How-to that works for me with tb2.0. with Centos 4.5. Other tb users have found it to work as well. http://cat24.net/Mondoarchive.html
If you run
updatedb
on your tb this will update the file index and you can then run
locate libpthread
to find the occurrences of libptread on your machine. On my system I find libpthread.so.0 as follows:
/lib/libpthread.so.0
This may help you track down the problem.
John, thanks for your response
Sat, 12/08/2007 - 2:47pm
I had gone through your site during my searching.
So, here's what I've done:
ran updatedb
locate lipbthread returns:
[root@asterisk1 log]# locate libpthread
/lib/tls/i486/libpthread-2.3.4.so
/lib/tls/i486/libpthread.so.0
/lib/tls/i686/libpthread-2.3.4.so
/lib/tls/i686/libpthread.so.0
/lib/tls/i586/libpthread-2.3.4.so
/lib/tls/i586/libpthread.so.0
/lib/libpthread-0.10.so
/lib/libpthread.so.0
/usr/lib/libpthread_nonshared.a
/usr/lib/libpthread.a
/usr/lib/libpthread.so
running mindi -
fails to create a floppy boot disk - kernel is too big; no surprise there.
it creates boot and data disk images.
backup to /server (a samba mount on my fileserver) goes fine.
trying a 'compare' still blow up - libpthread.so.0, and others all not found during a mount attempt.
So, now, how does this help me/what do I do now?
Mondorescue support site
Tue, 12/11/2007 - 5:46am
Here is the link for the Mondo Docs / Mailing list / Mailing list Archives and Wiki http://www.mondorescue.org/support.shtml
When I have run Mondorescue I set the verify option and looking at the Mondo logs at the end there don't seem to be errors with Centos 4.5
less /var/log/mondoarchive.log
It's during the restore that I bomb out
Tue, 12/11/2007 - 7:46am
The backup runs perfectly including the compare - no error messages, everyone seems happy.
When I boot the backup CD (and I need to use the ramddisk_blocksize=1024 command) and type:
interactive ramdisk_blocksize=1024 compare
The disk drops lots of text on the screen during the boot process and finally complains about not finding libc.so.6 and libpthread.so.0 and give me the option to type exit to reboot.
If I install Centos 5 with the 2.4 beta, I get a clean backup and a clean boot/compare (I haven't done a restore).
I'd really like to use 2.2.9, but I need a working backup to get myself up and running.
OK, I install 2.2.10
Tue, 12/11/2007 - 5:06pm
John,
Per your suggestion, I installed 2.2.10 - exactly the same issues on the restore/compare booting from the newly created CD.
My hardware is an Gigabyte SinXP1394 board with a zma800p and an LSILogic i4 RAID card.
All defaults are selected with Mindi.
I used the following commands to install mindi/mondo:
yum -y install cdrecord
yum install mkiofs
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/afio-2.4.7-1.i586.rpm
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/buffer-1.19-1.i386.rpm
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/mindi-busybox-1.2.2-2.rhel4.i586...
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/mindi-1.2.0-2.rhel4.i586.rpm
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/mondo-2.2.0-2.rhel4.i586.rpm
rpm -ivh ftp://ftp.mondorescue.org/rhel/4/mondo-doc-2.2.0-2.rhel4.noarch.r...
Am I doing something wrong?
Mondo and Centos 4.5
Thu, 01/17/2008 - 8:08am
I have posted a question on the mondo rescue forum in mid December
http://sourceforge.net/mailarchive/forum.php?thread_name=a659c1c0...
Someone else (not a trixer afaik) with the same issue has posted a similar query for Centos and Mondo
http://sourceforge.net/mailarchive/forum.php?thread_name=001c01c8...
As far as I can see the problem is to do with mindi-busybox and Centos 4.5. Busybox provides the normal shell commands in a compact way. Somehow busybox does not point to the correct libc and libpthread library locations. (although it did for Centos 4.4 when I wrote my how-to). I think that there are a couple of symlinks missing or pointing to the wrong target so that libc and libpthread cannot be found. I just have not had the time in the last month to sort out what is wrong.
I tried the latest Mondo version and then I got a kernel panic when I ran a compare. The seccond post above reports the same problem.
I think that the problem does not exist with Centos 5.
It shows that it is absolutely important after making a mondo archive iso image to boot the machine that has been backed up with the mondo iso disk and try a "compare". Since the verify option is always happy even when compare fails. If compare fails then nuke will fail also with the same errors.
When you do a compare which fails you are left with a bash shell prompt so you can run shell commands like cd, ls -al, mount, find, grep, etc........ and you can then try to diagnose the cause of the error.
I have not given up on this but I am short on time now. If any one else can look at the problem, and make some suggestions, it would be a great help, and I can then update my how-to.
Member Since:
2006-08-06