how to forcefully clean up EMM database?

[FlyMountain]

On DR isolated network, we recovered entire catalog. But we did see all volumes, devices and media servers which not existing in our DR enviornment. How can i manually clean them up from EMM?

I will still need meta data, image files for my application restores.

We are on 7.5.0.6 HP-UX IA64.

thanks

[mph999]

You could try nbdecommission command, but this might fail if the media server is not connectable.

Failing that, delete the volumes from NBU and use emmcmd -deletes -machines xxx -machinetype media to delete them.

Personally, I'd probably just deactivate them, effectively making NBU ignore them

See TECH60323 (sorry, on tablet and can't paste link)

[Marianne]

Next DR test, try partial catalog recovery and only restore images.

Add devices that are at DR site and inventory the robot to only add the media you brought from production site.
Add FORCE_RESTORE... entries to ensure DR server(s) will be used for restores.

*** EDIT ***

You stopped responiding to advice over here?
has catalog recovery method been changed on 7.5.0.6?

We have also suggested 'Recovery Without Import' over here:
can catalog files of a backup to be moved manually to different domain for restore? 

[FlyMountain]

thanks Marianne.

Actually the parital catalog recovery is alway our first option to recover our NBU master in our DR testing.

then we use "cat_export" and "cat_import" to merge backup meta data for all clients which we need to restore. But this time we have around 20 clients to restore. it does take long time to do it one by one and sometime the "cat_export" keep failing. this is why we decide to recover the entire catalog to make it simple. But we found we couldn't deactive the media servers which not existed, all confgiured tape libraries have been overwriten by "entire catalog" recovery.....

[FlyMountain]

thanks.

looks everything has dependencies and couldn't be reoverd from database. Should we clean up them by modifing database tables? I am after to touch the NBDB since it is so sensitive. the media servers also couldn't be deactived but they all marked as offline. we did see a lot of "bpcompatd" demons hanging around. looks all media servers attemp to connect.....

[mph999 · Accepted solution]

You just have to undo the dependencies one by one, eg, delete STU, then delete disk pool etc ... until you are just left with the media server, which will then delete.

Or, as I mentioned, just deactivate them, then it makes no difference.

[FlyMountain]

thanks. i believe you are right. we should undo all the dependencies before detele them accordingly.

Looks we are unable the deactive these media servers since they are marked offline....

[FlyMountain]

All,

I did capture another issue this morning.

During the "partial catalog recover" process, look the "image cleanup" job deleted a lot of un-expired images file. Our images files sized as 300GB currently, the restored job completed with same size, after "image cleanup" completed, the /usr/openv/netbackup/db/images has been resized to 20GB. Not sure what is going on?

Also i did noticed a mesage from "image cleanup" job, "importing header files into database". does it mena the "image cleanup"  is automatically doing "cat_export" and "cat_improt" jobs?

thanks

[mph999]

Have you tried, I think it should still work ....

[Marianne]

Bear in mind that you may have customized Retention Levels in Production.

Image Cleanup will apply Retention Levels at DR site.

Compare retention levels at the 2 sites with :
bpretlevel -U

You may also want to create NoExpire touch file at DR until restores are completed.

# touch /usr/openv/netbackup/bin/NOexpire on the master server.

[FlyMountain]

i can delete policies. some SLP couldn't be deleted with "images in process" warning. Since that, some STUs cannot be deleted. I am having issue to delete devices as well. I heard it may be caused by media servers offline. tried using "nbstlutil -cancel ... -force", it executed but the SLPs are still alive... ;(

[FlyMountain]

thanks for your tips. Looks recover "entire catalog" is better way and more reliable. Now i will need to find a way to clean up STU, devices.... i still have difficulty with these.

[Marianne]

You can try nbdecommission or nbemmcmd -deletealldevices.......  

See :

http://www.symantec.com/docs/HOWTO106701 

and

http://www.symantec.com/docs/TECH67656 

Partial recovery for DR purposes should be fine. Really!