make recovery question

[maria paschali]

hi,

If I use make_recovery -A -v -C -i -t "Titile Info". It writes from kernal information from my production system (A500 running HP UX 11.0 [2 processors]) to a DDS tape drive which is connected to the production system.

If I disconnected the tape drive from the production system and attached it to my development system (A400 running HP UX 11.0 [1 processor]) and boot the make_recovery tape for testing.

How does it distinguish between the different hardware on both machines? Ie the different processors, and different size and amount of disks?

I am just trying to understand how make recovery works (in terms of different systems).

Thanks for all your help in advance.

Maria

[Animesh Chakraborty]

Hi,
In your case you can use the make recovery tape as you are using same server model.

make_recovery calls make_medialif which according to the system?s architecture
copies one of 3 temporary kernels on the tape. VINSTALL is used for V class
systems, WINSTALL is used for N class systems, and INSTALL is used for other
systems. When the make_recovery tape created for a V class tries to modify the
kernel on the K class, the system will panic.

To properly bootup from the make_recovery tape, the following steps need to be
performed.

1. Create the make_recovery tape for the specific architecture; see the
make_recovery(1M) man page for more details.
2. Insert the recovery tape into the tape drive.
3. Boot the system.
4. Interrupt the boot sequence to redirect it to the tape drive, specifying
the make_recovery temporary kernel at the end of the line. For example
boot [scsi_addr_tape_drive] VINSTALL, if the make_recovery tape is
created for a V class.
5. Elect no interaction with ISL.
6. Allow the install process to complete automatically.

http://us-support.external.hp.com/cki/bin/doc.pl/sid=91c21ae10541310a2a/screen=ckiDisplayDocument?docId=200000046942472


Regards
Animesh

[Sridhar Bhaskarla · Accepted solution]

Maria,

It's the same way a running system recognizes the hardware changes. For ex., if you take out a processor or add a card to a running system, the kernel is going to identify it. Similarly when the make_recovery boots through the image,it loads the kernel that by virtue of it keeps track of inventory of the system and adds (only) the instance files through the normal insf procedure.

However, all the files are recovered by extracting the tar archive of make_recovery archive. So, the backedup kernel /stand/vmunix would also be recovered. So,if the recovered system has a new type of card, it will remain UNKNOWN to the system. Also as an example, if the previous system has a boot disk at a certain path with a device file say c0t0d0 and if the new system has the bootdisk at a different path, make_recovery is going to create a different device file for this disk. However,it is going to recover c0t0d0 also from the archive and ofcourse doing an lssf on c0t0d0 will give out???.

-Sri

[maria paschali]

Hi,

thanks for that.
It has helped.

Where can I find more info on it? I want to familarise myself a bit more in the workings of such things.

thanks again
maria

[Michael Tully]

Hi Maria,

Have a look at the following Ignite document.

http://www.docs.hp.com/hpux/onlinedocs/B2355-90704/B2355-90704.html

HTH
-Michael