It's been a very, very long week, but we have successfully restored the ADSM
database from a VMBackup tape. Although the process was agonizingly slow, it
was straightforward, easy to do, and highly successful. Working pretty much
non-stop (the ADSM salvage commands were working non-stop) it took almost
exaclty a week to restore a 40,000,000 million record database and audit all
the stgpools.
For anyone who might want to keep this stored away as a last resort in the
event of disaster, the steps we took were:
1. Restore the weekly vmbackup 2. Restore the vmbackup incrementals.
3. Do a dumpdb 4. Do an install
5. Do a reloaddb 6. Do and auditdb with fix=yes
7. Halt the server 8. Add NOMIGRECL to the options file
9. Change Expinterval to 0 10. Restarted the server
11. Disabled the server 12.Did an audit vol on the stgpools
13. Halted the server 14. Put the options file back the way it was.
15. Started the server back up
None of these steps was difficult, just slow. We didn't have any surprises
and once a command was initiated it just kept chugging along until it was
finished. Hats off to the ADSM development team. We have since instituted
the ADSM backup routine, but it gives us even a warmer, fuzzier feeling to
know that if all else should fail we STILL have another backup that can be
used. And huge thanks to all of you who helped us with this with
recommendations and encouragement.
|