[Bacula-devel] Re: [Bacula-users] Update: Problem in using Bacula Rescue CD feature

[Kern S. <ke...@si...>]

On Mon, 2004-09-06 at 17:45, guarav gupta wrote:
> Hi Kern
> 
> I resolved the issue . Currently the restore for bare
> metal recovery is going on on my system. Lets hope it
> works without errors.
> Following are the issues i identified in the
> getdiskinfo script.
> 
> 1. There were two lines in the script
> mount -l -t ext2 >mount.ext2.bsi
> mount -l -t ext3 >mount.ext3.bsi
> 
> I added the following line in it to support reiserfs
> file system (was testing on Suse)
> mount -l -t reiserfs >mount.rei.bsi

Yes, good idea. I've now added that to the script.

> 
> 2. In the following lines (for generating
> start_network script
> host=`hostname`
> ip=`host $host | cut -d ' ' -f 4`
> if [ $ip = "out;" ] ; then
>   ip=`ifconfig | grep inet | head -1 | sed -n 's/\
> \+inet addr:\([0-9]\+\(\.[0-9]\+\)\{3\}\).*/\1/p'`
> fi
> 
> I had to comment the following two lines.
> ip=`host $host | cut -d ' ' -f 4`
> if [ $ip = "out;" ] ; then
> 
> as in the start_network file i was getting "ip=found",
> because `host $host | cut -d ' ' -f 4` was retrieving
> 'found' on Suse. On redhat it was working fine.

I've done this a different way, so it should now correctly detect what
is currently running on your system for eth0 and eth1 if either or both
are up. It will then reset them to the correct device.

What I haven't yet done is to correctly parse and reset the routes
defined. It simply sets eth0 to be your default gateway -- that should
work fine in 99% of the cases.

> 
> 
> 3. Inorder to support Reiserfs i had to add the
> following line in the getdiskinfo(for generating
> correct mount_partitions script on Suse)
> 
>  sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mkdir -p
> \/mnt\/disk\2/p' diskinfo/mount.rei.bsi >>mount_drives
>  sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mount \1
> \/mnt\/disk\2/p' diskinfo/mount.rei.bsi >>/tmp/1$$

Yes, thanks again. I added that too.

> 
> 
>  4.
> 
>  Inorder to generate the correct partition.hda i had
> to change (In getdiskinfo)
> 
>  "sfdisk $i <$di/sfdisk.make.$j.bsi | less"
>  to
>  "sfdisk $i <$di/sfdisk.make.$j.bsi"

I don't understand this. Is less not on your system and not in the /sbin
after rebooting? The only purpose of less is to allow you to examine
what sfdisk is doing. I hesitate to remove it because it is very useful
and should harm nothing.

> 
> 
>  5. Also the issue of more than one ethernet card
> needs to be handled. I have to manually switch the
> ethernet jack to make it work. Please fix the issue ,

I think I've fixed this problem so that both cards should be brought up
with the correct IP addresses. There may be a problem if you are using
eth1 as your default gateway or if you have some complicated routes.

If they are not brought up correctly, please send me the output from
your normal system (not the rescue system) to:

  ifconfig
and
  route

> I would love to test it again. If i find anything
> about it, i will try to do it.

If you can pull the latest CVS, please do so as it contains the fixes I
mentioned above. It is version 1.35.3 06Sept04.  I am most interested in
knowing if I fixed the network start problems.

Thanks again for testing.

Best regards, Kern

PS: did you figure out why there were permission problems while
restoring?