Disk Staging - Advice on Restores?

[GWatson]

Hi,

Firstly, I am running Netbackup 5.1 MP3 in a Windows 2003 Server environment.

I am currently testing disk staging and having a few problems with restores\getting my head around it.

Currently we keep our tapes on a 4 weekly rotation, however the disk copy will only be there for one week due to the size of the NAS. I simulated a restore job by deleting the disk copy thinking/hoping it would resort to tape, but it failed. See below:

"29/11/2005 15:59:23 - begin Restore
29/11/2005 15:59:23 - restoring image uk****_1133191985
29/11/2005 15:59:24 - connecting
29/11/2005 15:59:29 - connected; connect time: 00:00:05
29/11/2005 15:59:29 - started process bpdm (5844)
29/11/2005 15:59:29 - Error bpdm(pid=5844) cannot open file E:\Full\uk01-w888_1133191985_C1_F1, No such file or directory
29/11/2005 15:59:34 - restored image uk****_1133191985 - (media open error(83)); restore time 00:00:11
29/11/2005 15:59:36 - Warning bprd(pid=6076) Restore must be resumed prior to first image expiration on 29/11/2005 15:33:05
29/11/2005 15:59:36 - end Restore; elapsed time: 00:00:13
the restore failed to recover the requested files(5)
29/11/2005 15:59:39 - Error bpbrm(pid=3192) client restore EXIT STATUS 83: media open error"

With my current setup does it mean the tapes which are kept offsite are useless after orginal backup is ovewritten, surely this defeats the point in disk staging?

Thanks in advance

Gareme

 

[GWatson]

Just tried it again on a file which hadn't expired....


16:47:21 29/11/2005: Restore Started

16:47:21 (17330.001) Restoring from image created 29/11/2005 04:06:09
16:47:26 (17330.001) INF - TAR STARTED
16:47:31 (17330.001) ERR - The following files/folders were not restored:
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/ACNTTYPE.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/IATSYSPA.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/MODEM._T_
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/PRINTER.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/RESPCODE._T_
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SERVICE.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SRVAUTHL.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SRVFRACC.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SYSDETLS.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SYSMESG._T_
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/SYSNOT.___
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/TRANSDEF._T_
16:47:31 (17330.001) UTF - /D/Barclays/V9Softwareold/DATA/UKPAPACC.___
16:47:31 (17330.001) Status of restore from image created 29/11/2005 04:06:09 = media open error

16:47:31 (17330.001) INF - TAR EXITING WITH STATUS = 174
16:47:31 (17330.001) INF - TAR RESTORED 0 OF 0 FILES SUCCESSFULLY
16:47:31 (17330.001) INF - TAR KEPT 0 EXISTING FILES
16:47:31 (17330.001) INF - TAR PARTIALLY RESTORED 0 FILES
16:47:33 (17330.xxx) INF - Status = the restore failed to recover the requested files.

 

[comtec17]

Are you sure that the image had successfully been migrated to tape before you deleted the file?

Have you promoted the copy on tape to the primary copy?

 

[syjl]

comtec is exactly right. you must promote the tape to primary copy before the restore. you should never delete disk staging images unless they are not in the images directory. You want to restore from disk whenever possible. There is a tech note on cleaning up disk staging for images that are caught in no man's land. I have a script that helps identify those images that need to be deleted but it needs somes work before it's perfect.

 

[GWatson]

Thanks guys

"Are you sure that the image had successfully been migrated to tape before you deleted the file? "

Yeah, I can see the image on the "Media Contents" report.

"you must promote the tape to primary copy before the restore."

Ah ok. However will this mean the backup goes to tape first then to disk, that's what I want to aviod. Also where do you set this?

"You want to restore from disk whenever possible."
Yeah I realise this, but we simply don't have the capacity to keep all fulls on disk. Our current full will always be on disk, but we need tape for historical reasons, like everyone else I guess.

Graeme

 

[GWatson]

hmmm, found this on the Veritas help

"The image exists on both the disk staging unit as well as the final destination storage unit until the disk staging storage unit is full. The image on the disk storage unit continues to be the primary image until it is expired or removed, after which the image on the final storage unit is the primary image."

Sounds like when the image is deleted off disk (which is what I've done) it should automatically make the tape the primary source... strange...

 

[comtec17]

That technote is if the image was removed by NetBackup, not deleted by the user. You must first expire the image that is on disk using the bpexdate command for the image on tape to become the primary copy.

 

[GWatson]

Thanks for clearning that up.

So when the next weekend comes around it will automatically remove the old image to make room for the new backup then promote the tape to the primary. So basically I only need to do something when I manaually delete an image.

Graeme

 

[comtec17]

correct.

 

[tech2004]

I have seen issues where once the NAS or SAN space is full, the oldest images still didnt get deleted causing new jobs to fail. As mentioned earlier, once an image has been copied to tape you can use bpexpdate to make the copy on disk the secondary and the copy on tape the primary.

Below is the command to run on your NBU server to expire the disk images
which will free up space on your array. First, use Windows to browse to your
disk stage. Locate the files with extention .ds - these files are created
automatically once NBU copies a disk stage image to tape. You'll notice the
name of the .ds file is something like: servername_1102823822_C1_F1.ds. Copy
the servername_1102823822 part of the file name. Open a command prompt and
change directories to c:\program files\veritas\netbackup\bin\admincmd and
type:

bpexpdate -backupid servername_1102823822 -d 0 -copy 1

Again, only use the above command if using DISK STAGING.
This will expire this backup image and automatically delete the files
associated with it from your hard drives.

 

[GWatson]

Excellent, thanks very much for that.

We did our first disk staging at the weekend around 1.8TB. It worked a treat, although a little slow copying it to tape, took almost 48 hours. But that's because it was using one tape drive. I found a solution which may be of benefit to others.

http://seer.support.veritas.com/docs/278169.htm

Thanks again for you help guys.

Graeme

 

[GWatson]

Just waiting for MP1 to be released before I install version 6 on advice from our pre sales Veritas guy.