Re: Marking a DBspace as down

viegasjp@panasonic.com (Viegas, John Paul) wrote: > >Hello, > We are trying to test a single dbspace restore with ONBAR, but we are >having trouble marking that dbspace as down. We are using raw > partitions, or else we would just delete the dbspace file in the >appropriate file system. If anyone has been successful in bringing a >DBspace down your help would be greatly appreciated. > >Thanks >JP > Here's a method that was posted some time ago and it does work. 1. Perform system backup:- onbar -b etc. 2. Corrupt chunk:- e.g. dd if=rubbish_file of=/dev/chunk1 bs=512 conv=sync 3. Take system offline:- onmode -ky 4. Bring system back online:- oninit (online log should indicate a chunk down) 5. Set environment variable:- export ARCHIVE_TEST=true 6. Find out the chunk number of the down chunk:- onstat -d (flags ND for dbspace and PD for chunk - note dbspace shouldnot be flagged as down yet, just the chunk) 7. Foce related dbspace down:- onmode -o <chunk number> 8. Check that dbspace is now flagged as down:- onstat -d 9. Allow system to checkpoint with down dbspaces:- onmode -O 10. Initiate warm recovery of dbspace:- onbar -r chunk1 11.Backup current logical log:- onbar -l -c (this is very important - do not skip - fake the backup if you are not backing up logical logs) Please note that this procedure is my own work and is not supported by my employer - use it at your own risk, although it works fine for me. Any questions, please e-mail me at danny@uk.ibm.com as I do not read this newsgroup - just found it when I was doing a web search. Hope this helps.