Re: [Bacula-devel] potential error in rescue scripts

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

Hello,

On Tue, 2004-09-14 at 13:59, philip nash wrote:
> --- Kern Sibbald <ke...@si...> wrote:
> 
> > Hello,
> > 
> > On Fri, 2004-09-10 at 14:07, philip nash wrote:
> > > Hi all
> > > 
> > > I was just going through the bare metal restore
> > rescue
> > > scripts. I couldnt get the logic of the following
> > > lines
> > > 
> > > **********
> > > sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mkdir -p
> > > \/mnt\/disk\2/p' $di/mount.ext2.bsi >>
> > mount_drives
> > > 
> > > sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mkdir -p
> > > \/mnt\/disk\2/p' $di/mount.ext3.bsi >>
> > mount_drives
> > >      echo "#" >>mount_drives
> > > 
> > >  sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mount \1
> > > \/mnt\/disk\2/p' $di/mount.ext2.bsi >/tmp/1$$
> > > 
> > > sed -n 's/\(^.*\)\ on\ \(.*\)\ type.*$/mount \1
> > > \/mnt\/disk\2/p' $di/mount.ext3.bsi >>/tmp/1$$
> > > 
> > > # sort so that root is mounted first
> > > 
> > > sort -k 3 </tmp/1$$ >> mount_drives
> > > *********
> > > 
> > >  As I see it, if the entire directory hierarchy
> > > (/mnt/disk, /mnt/disk/prog, ... )is created first,
> > and
> > > the partitions are mounted with root (/)at the
> > top,
> > > the underlying directories will become invisible
> > and
> > > the remaining stuff wont work.
> > > 
> > >  I suggest it should be like:
> > > 1. create the directory (/mnt/disk) for the root
> > > parttion
> > > 2. mount it
> > > 3. create the remaining directories
> > > 4. mount them in a similar fashion
> > > 
> > >  What is your opinion on this?
> > 
> > If you think there is a problem here, you will need
> > to show me a real
> > case where it fails. Please send a "df" of your
> > system, and the output
> > file that getdiskinfo produced along with an
> > explanation of why it won't
> > work. You could also possibly attach a few of the
> > files that are
> > processed so that I can repeat it here with your
> > data. 
> > 
> > Please put the above into the bugs database so that
> > it can be properly
> > tracked and processed.
> > 
> > Regards, Kern
> > 
> > > 
> > > Regards,
> > > Philip
> 
> Hi everyone,
>  I apologize for the delay. Got caught up in some
> chores and hence the late reply.
> 
> The df on my system gives:
> [root:root]# df
> Filesystem 1K-blocks      Used Available Use% Mounted
> on
> /dev/hda1  4195632   4083892    111740   98%    /
> /dev/hda3  20968736  19650192   1318544  94%    /boot
> /dev/hda4  12639928   8464084   4175844  67%    /prog
> tmpfs      512460         0     512460    0%   
> /dev/shm
> 
> The file mount_drives contains lines like (I didnt
> save the old file and have changed the generating
> script now, so these aren't the *exact* lines)
> 
> mkdir -p /mnt/disk
> mkdir -p /mnt/disk/boot
> mkdir -p /mnt/disk/prog
> 
> mount /dev/hda1 /mnt/disk
> mount /dev/hda3 /mnt/disk/boot
> mount /dev/hda4 /mnt/disk/prog
> 
> Now, when this is executed, the last 2 mounts fail
> because the target directories do not exist. I suggest
> that the order should be:
> 
> mkdir -p /mnt/disk
> mount /dev/hda1 /mnt/disk
> 
> mkdir -p /mnt/disk/boot
> mount /dev/hda3 /mnt/disk/boot
> 
> mkdir -p /mnt/disk/prog
> mount /dev/hda4 /mnt/disk/prog

Well, I must be really missing something here because I don't see
anything wrong with the first script you show above. It should create
the mount points with no problems.

That said, I know that the rescue scripts have some difficulties getting
the order of creating directories correct. I you can send me something
it does wrong, and I can understand why it fails (I don't understand why
the first script above fails) I'll try to fix it.

Regards, Kern

> 
> also, I have one question: in the various mount
> outputs that I have observed, the root partition (/)
> always comes first. Can we depend on this behaviour or
> is it just coincidence?
> 
> Hope someone clarifies things.
> Regards, 
> Philip
> 
> 
> 		
> _______________________________
> Do you Yahoo!?
> Declare Yourself - Register online to vote today!
> http://vote.yahoo.com
> 
> 
> -------------------------------------------------------
> This SF.Net email is sponsored by: YOU BE THE JUDGE. Be one of 170
> Project Admins to receive an Apple iPod Mini FREE for your judgement on
> who ports your project to Linux PPC the best. Sponsored by IBM. 
> Deadline: Sept. 13. Go here: http://sf.net/ppc_contest.php
> _______________________________________________
> Bacula-devel mailing list
> Bac...@li...
> https://lists.sourceforge.net/lists/listinfo/bacula-devel