Ignite-UX at DR site using VPARS

[Susan Pellerito]

Hello.
We are planning a DR test at our contracted remote DR site. What I need to recover is 2 rp7410's to 2 N4000 servers. At home, these are both the primary vpars in a multi vpar environment. In my previous test last year, I was successful in recovering one of the servers using make_recovery. The other crashed while booting from the ignite tape. I attempted another tape and the same thing happened. I was forced to rebuild from scratch. This time around I would like to feel confident the ignite tape will boot as expected. I will have no need to rebuild the 2nd vpar on either box. Therefore, can I simply reboot my server into a non-vpars server and run make_tape_recovery. At the DR site, I will recover from this tape, use vparmodify to point to new hardware for vpar1 and then boot vpar1. What if I don't even boot the vpar1? Will the primary vpar contain the kernel parms and users of the primary vpar? I would test this, however, it is a production server and not available to take down. Any help/suggestions appreciated.

[Steven E. Protter]

Shalom Susan,

Its hard to predict the outcome without knowing what failed the last time.

IT could have been anything from a missing patch on a vpar to a hardware incompatiability.

What was the difference between the good server and the bad server?

To be exact, you should be able to at least run make_tape_recovery or make_net_recovery on every vpar. Then put the vpar structure back together and use the Ignite images.

Will the primary vpar contain the kernel parms and users of the primary vpar?

The answer depends on how you backed up.

I'd take a number of possible backup's with you to the DR site and leave time to test them all. You can't bring down production for a test, thats why you are going into the DR test in the firs place.

I'm not sure booting the server into non-vpar mode and doing the make_tape_recovery will work. But I think its worth a try. Ignite needs a running system with all hardware allocated and available to do a good, usable backup.

I think your strategy is sound. If you also have recovery images, made in the running vpars, you have covered all your bases and will recieve your answer at the DR center.

I know, it not satisfying, but you're going to the DR center anyway, why not test both scenarios.

:-)

SEP

[Susan Pellerito]

Thanks for your comments. I appreciate your statements regarding the reason I am in the DR test site is to try different scenerios. However, in our case, we have a limited time frame and it is expected that I not only get the server back up and functional, but configure and restore the Oracle dbase and application server using 3rd party (Netbackup). Therefore, I am under the gun to get the OS up asap--not the best strategy, but there it is. Anyway, I was wondering exactly you meant by the following:
"Then put the vpar structure back together and use the Ignite images."

[Steve Faidley]

I believe if you run a make_recovery and restore from vparX, that is all you have to do other then import your non root VG's.
When you boot one of your vpars directly (ISL hpux rather then VPMON) you see all hardware.

It is hard to say why you crashed last year.
Were both N4000's at the same firmware level?
Did your kernel parms "fit" on the DR server?
Are the second servers make_recovery tapes any good?
Can the make_recovery tape be restored anywhere, (like on the server that did restore)?

You may want to add time to attempt to restore any failed make_recovery tapes to an alternate server(that did sucessfully restore) during your DR test.

[melvyn burnard]

If you create an Ignite image from a vpar, and then restore this to another server, then boot it into Npar mode, you will still have the same OS image, just not using Vpars.

The concern is why you had th eorigional issue, and I suspect as mentioned by others that there was either a patch missing, or more likely a hardware/driver/firmware issue.
Have the DR site servers been checked to make sure they have the relevant firmware installed?

[Susan Pellerito]

Thanks for all your comments. As the DR site is in New Jersey and I am working remotely from a console in LA, my visibility into the DR hardware is limited. Because the DR site was pointing towards the VPAR situation, I was following their lead. This time, if there is a problem I will be more suspect of the hardware. Several of you mention the make_recovery vs make_tape_recovery. Yesterday I made the make_tape_recovery tapes of the running vpars. Should I use make_recovery instead?