The question was:
> We are preparing our "How to get out of troubles quick "
> scenario.
>
> System: SS1000, Sun DAT, Oracle, Legato Networker, Solaris 2.3,
> Server runs Legato by itself.
>
> Worst case scenario: We are loosing internal drive with
> root partition & Legato software.
>
And general opinion was:
1) if it is possible, create a second bood disk and on a regular
basis, copy root + /nsr partiton to backup drive;
2) if it is not possible, then I will have to do all yoga with
booting from CDrom, doing miniinstall, and restore from tapes.
3) Since we do have Disk suite, then option of mirrowing
/ /usr /var to second drive was great..
Thanks to:
From: sven@mpim-bonn.mpg.de (Sven Maurmann)
From: bharrell@digit.com (Bruce L. Harrell)
From: tkevans@eplrx7.es.duPont.com (Tim Evans)
From: glenn@uniq.com.au (Glenn Satchell - Uniq Professional Services)
From: rgjj490@wadnr.gov (rocky gould)
From: don@doug.med.utah.edu (Don Baune 581-6088 MIRL)
From: cygan@wpm.com (Linda Cygan)
From: Anchi Zhang <anchi@Starbase.NeoSoft.COM>
_zjt ( wannabe in training )
Here are excerpts from letters:
( full text is available for FTP as :
ftp.wariat.org:/pub/backup
)
In principle you are right with your idea of dumping the internal
disk to another one (or to a DAT).
I would suggest to make a complete dump of the internal disk (preferable
via mirroring to another disk) and just swap this disk in.
ATTENTION: To restore the root partition it does not suffice to just bring
the dump image back to disk, but you have to run "installboot" afterwards
to make booting possible (see the man pages ...).
>>>
Is this your legato server you are doing this to?
Keep a copy of your fstab file. You will need this in order to restore
properly. (partion correctly)
>>>
I'm not clear that Legato's backups are readable with standard
Sun utilities such as 'ufsrestore' or 'tar'. Even if they
are, be careful not to restore 'ufsrestore' itself with itself.
You'll overwrite it and mangle it.
>>>
Actually you can get away with doing a one-off dump of / and /usr to
DAT. In the event of failure you restore that, then use Networker to
restore the latest /, /usr and all other partitions. You could do dumps
of / and /usr only when you do major changes such as new configuration
or OS release, etc.
>>>
My big concern would be the Legato indexes and what kind of disaster recovery
mechanisms are required to bring it back up to date.
Other question would be that daily dumps of / and /usr would seem futil as
those two partitions should rarely fluctuate ( especially if you take care of
tmp ).
>>>
Why not put the second drive on the same machine? Then you
can create scripts to keep it current to the hour or even
min. This disk would be a boot disk also. If the first one
dies then either boot manually from the second address or change
the address to be the same as the first.
Even better yet Disk Suite can mirror the drives for you and then
when you get a failure every thing is in sync.
>>>
1st you have to obtain another drive. If this takes any
amount of time the restore time becomes a very minor part of the
down time.
>>>
One thing to remember with Legato, if you have a lot of data to restore,
using the command line commands rather than the GUI is much much faster.
I start a restore using the recover command in a shell-tool and
then run a nsrwatch on another machine to make sure I have the right
tapes loaded and know when it needs another tape.
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:09:13 CDT