[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: recreating lenny



Ross Boylan wrote:
> I think I need to recreate a lenny system, and could use some advice
> about how to do it.  The system was i386, while the one I have now
> is amd64 (wheezy, though I have access to an amd65 running lenny).

You can use debootstrap to create a lenny chroot.  I just tried this
and it is still working fine for me.

  debootstrap lenny lenny-dir http://archive.debian.org/debian

Tbere is additional hints here:

  http://wiki.debian.org/chroot

If you have difficulty please feel free to ask more questions about
getting the database daemons running in the chroot.

Alternatively, you could install an old system on a different piece of
hardware.  Just put in a temporary disk into any machine and install
the old system and then walk it forward through the upgrades.  The
installation disk for Lenny 5 is here:

  http://www.debian.org/releases/lenny/debian-installer/

> I don't think the current debootstap can do lenny,

It can.  See the debootstrap line above.

> and using the lenny installer or (if I could get it) lenny
> debootsrap have hardwired assumptions about where to look for files.
> I don't know if there is a way to repoint them at the archive.

There are several different ways to install.  Assuming you use CD#1
then just avoid using a network mirror or security server.  If using
the netinst iso image to install from then answer all of the questions
as you would normally answer them up through the "Debian archive
mirror country" question.  At that point page up all of the way to the
top and select "enter information manually".  For Debian archive
mirror hostname enter "archive.debian.org".  Then the default
"/debian" for the archive mirror directory.  Continue forward.  Ignore
and continue through the warning about no security repository being
available.  Proceed normally from there.

> There are also chroots, VM's, and multiarch, but I can't see clearly how to 
> get those started.

Yes.  Lots of options.

I think debootstrap would be the easiest for you since you could do it
on your latest running amd64 system.  It is just a directory of files
that way.  Then recover the data and move it to your newer system.  If
you select another option that is fine too though.

> Background: My 32 bit lenny system died along with the primary hard disk.  I 
> have been able to restore the files from backup, but it seems in some cases 
> (e.g., databases) migration requires the tools of the original system (e.g., 
> to convert to an export format).

You have made me (and I am sure many others) very happy by saying
those magic words that you have a backup.  That is the way it should
be but something rarely heard.

> The backups were made from snapshots with the databases shut down; perhaps I 
> should have made the export dumps as part of that process.

Uhmm...  Yes.  That would have made things easier.  But you should
still have enough to be able to recover.  Should work out okay in the
end.  But would have been easier if you had a cron task in
/etc/cron.daily/local-mysql with something like this:

#!/bin/sh
umask 077
mysqldump --defaults-file=/etc/mysql/debian.cnf --all-databases \
  | gzip > /var/backups/mysql.dump
savelog -q -d -l -C /var/backups/mysql.dump

That would produce a series of files in /var/backups like this:

  -rw------- 1 root root   10321316 2013-04-24 03:30 mysql.dump.20130424033037
  -rw------- 1 root root   10321123 2013-04-25 03:30 mysql.dump.20130425033035
  -rw------- 1 root root   10327392 2013-04-26 03:30 mysql.dump.20130426033035
  -rw------- 1 root root   10328613 2013-04-27 03:30 mysql.dump.20130427033036
  -rw------- 1 root root   10330227 2013-04-28 03:30 mysql.dump.20130428033034
  -rw------- 1 root root   10327932 2013-04-29 03:30 mysql.dump.20130429033034
  -rw------- 1 root root   10332096 2013-04-30 03:30 mysql.dump.20130430033042

I am doing --all-databases because then I know I always have
everything.  But that makes recovery somewhat more difficult because I
usually need to do the recovery twice.  Once on a scratch system and
then recover just the individual database I need as opposed to
everything.  Doing each database in turn would be easier.  But I would
either need to script something or remember to update the file when I
make a new one and it would be worse if I missed one.

> P.S.  The backups are not complete disk images; e.g. they don't
> include /usr.  So I can't simply recreate the disks and start a
> chroot or VM.

Oh well.  You will have debootstrap'd yourself to a new one here in a
few moments anyway.  :-)

Bob

Attachment: signature.asc
Description: Digital signature


Reply to: