Hi James,

This is a cinch if you have VM (which you don't).  The nice thing about VM
is that even though you are running on different hardware at different real
addresses, your Linux guests don't know that - everything is still at the
correct virtual addresses.

Backing up your Linux system (using DDR or the like) is simple enough, and
restoring it will be simple enough - the problem will be (as you point out)
that the hardware will not be at expected addresses and fstab and other
things will fail.

Maybe you could have your DR people bring up a "generic" Linux image in the
DR LPAR, set up some basic networking, restore your backed up linux to
available DASD mounted to the "generic" Linux image and then you can tweak
it from there.  Complicated, definitely.

Michael Coffin, VM Systems Programmer
Internal Revenue Service - Room 6527
1111 Constitution Avenue, N.W.
Washington, D.C.  20224

Voice: (202) 927-4188   FAX:  (202) 622-3123
[EMAIL PROTECTED]



-----Original Message-----
From: James Melin [mailto:[EMAIL PROTECTED]]
Sent: Thursday, February 06, 2003 9:36 AM
To: [EMAIL PROTECTED]
Subject: Disaster Recovery scenario - comments, suggestions, real world
advice


We are having a Disaster Recovery exercise in April. My boss thinks it would
be nifty if we could recover the 2.4 Linux system I just got installed (with
much help from you SMOONOL's - Secret Masters Of Obscure kNowledge Of
Linux).

The disaster recovery exercise will not allow us to have an organization
employee with access to the HMC on which we will be operating, we would have
to rely on someone at the vendor site to do any HMC interactions for us. We
will be doing this in LPAR mode - We do not have VM, and with a 25% budget
cut across IT, we won't be getting it any time soon.

Secondly, the DR exercise will have us restoring OS/390 volumes to different
device numbers than what exist here. We have procedures in place for that
and it all works. I will have to do the same with the CDL backups of the
Linux volumes

since /proc/dasd/devices in any backup will reflect the old device mapping,
I assume I will simply fall flat on my face if I try to IPL from the CDL
copies.

So what is necessary after a CDL restore is done to new DASD in order to IPL
the Linux system? Is there an initrd that must be booted from the HMC so
that adjustments to the image may be done? Is there some OTHER way to
alter/access the CDL volume structure from OS/390 to make the changes
necessary to IPL? If we have to have the vendor operator performing initrd
boots and entering in the initial ramdisk information so that I could telnet
into the system and fix things that way.

Note, we will ONLY have 3270 dumb terminals at this site, from what I
understand. Contract has not previously included network attached desktop
PC's from which I can run a decent VT100 capable telnet session, or TN3270.
That alone may kill this.

So now that I've outlined this fairly hopeless situation.... is there
anything that can be done to allow us to accomplish a disaster recovery test
of Linux in the described environment?

Reply via email to