Disaster Recovery Using Old System's Disks

[Glenn Mitchell_2]

Providing a disaster does not destroy the disks in my Production database server (HP 9000/800 N4000 running HP-UX 11.11), would it be advisable to un-rack root disks and the SureStore E SC10 enclosed data disks from the affected server and transport them to another server loaded with the same version OS and vgimport each disk and volume group (as described in the previous thread) to implement a recovery – or is it easier to just do a tape recover

[Mel Burslan]

As the disks are more susceptible to damage while being in transit, it is advisable to have a tape copy of bootable tapes but for faster recovery, the disk option should work technically speaking but you do not want to get caught with pants down {grin} should the disks get damaged somehow.

At the time of diseaster recovery, the worst thing to figure out is that you do not have any bootable media. I Usually keep 2 copies of a BBT before attempting any drastic change to my systems.

[Steven E. Protter]

Tape recovery is better.

There are lots of factors that can go wrong on the disks. lif trouble, instance numbers, lvm structures.

make_tape_recovery and make_net_recovery to a an Ignite image server is a much better way to go.

SEP

[A. Clay Stephenson]

I don't think I'm talented enough to limit the scope of my disasters to just a few components and spare the critical disks -- maybe a Force Field around the disks would work. In your case, a tape recovery makes much more sense. You use Ignite to get your vg00 back together and then conventional backups can take over from that point. Moreover, it's a little difficult to test disaster recovery using production disks.

[Glenn Mitchell_2]

Thank you for your replies. I will plan on a tape recovery solution. My Production server is using LVM mirror disk but the server we will be using as a disaster recovery platform with RAID-5 (and appropriate disk space for OS and data). The DR site will be loading HP-UX 11.11 on the recovery system prior to our arrival at the site. Do I even need ignite in this case (with dissimilar systems and the OS already loaded? ... Or should I still bring a ignite make-recovery tape? Can I (or SHOULD I) use the ignite tape to restore "parts" of the OS - or am I asking for trouble?

[lawrenzo]

Yes you will require ignite as this will contain your original system files such as the kernel and other settings for networking etc.

You should also have a full backup of the system being recovered.

Also it is work noting the original filesystem configurations.

There should be a lot of planning before DR as it is too late when you arrive onsite.

hope this helps.

[Con O'Kelly]

Hi Glenn

If your DR platform is running on a different hardware platform then you may have trouble using an Ignite tape. Generally Ignite only restores to similar hardware platforms.

It sounds like you are building a replica DR system to be used in the event of a disaster.
You could try using Ignite to build the server. If this doesn't work then you'll need to install 11i OS and restore necessary files/directories from a regular backup (fbackup,cpio etc) of production server. This is much easier than trying to restore files from an Ignite backup.
Also ensure you install all necessary patches etc so that this erver mirrors the production server.
Ensure that you the build the DR server with same LVM setup etc.

Finally you'd want to test that you can get the applications up and running on the DR server.

I'd still do regular ignite backups of the production system as this are very useful for quickly restoring a system that has a major system crash.

Hope that helps.

Cheers
Con

[A. Clay Stephenson]

I only trust Ignite to get me back if the target is essentially identical; it may work just fine otherwise but I never will put myself in the position of may work when it applies to disaster recovery. I wouldn't even trust anyone to be able to install HP-UX. If this were my DR plan using your scenario part of my offsite media storage would include KNOWN GOOD copies of the OS Install Media + codewords + Application CD's + QPK's. You also have to keep copies of your kernel tunings and disk layouts.

If this were me, I would maintain a swdepot as part of my regular backup and include in it all installed software and patches. It is then easy to restore this depot and swinstall from it to get all your software and patches back on the box.

[Glenn Mitchell_2]

I'm hearing a lot of good information in your replies. Thank you! I will have to re-evaluate my present tape backups to include the necessary patches and sw depot. I'll also include the present kernel parms and disk & filesystem layout diags as well in my Sys.Admin. binder (for ). Because the systems are so dissimilar, I will also rely more on doing a full OS build, filesystem rebuild - and less on the ignite option (although I'll continue to keep my ignite tapes up to date).
Thank you all again!