upgrading to Netbackup 6 'killed' all my pools & my Catalog

[aske_andersen]

Hi

I've just upgraded our Netbackup 5.1 -> 6.0 -> 6.0MP3
(running on Win2000-SP4 server).

When I start the admin. console,
the Catalog is empty and all my pools are gone.
(all my policies survived )

I've tryed to import my catalog back into Netbackup via the 'Initialize import' under Catalog. No luck.

I have the latest 5.1 catalog on a tape AND on the HD.
the catalog folder contains 4 files:
Header
Image1
Image2
Image3


First:
How do I import my catalog to v6.0 format/use ?

Second:
How do I get my pools back ?

Regards...

[DavidParker]

Aske,
To re-import your catalog, you will need to use the bprecover command, not the Import.
You can find bprecover here: \VERITAS\NetBackup\bin\admincmd\

I think Volume Pools are part of the catalog backup (\volmgr\database); if you restore the entire catalog, they should come back.

Hope that helps!

Oh, just in case, you may find this note helpful:
http://seer.support.veritas.com/docs/282327.htm

More useful information:
http://seer.support.veritas.com/docs/285223.htm

[Dennis_Strom]

did you run the recover command as per the instructions?
A simul post dave beat me to it.Message was edited by:
Dennis Strom

[DavidParker]

Ah, but you remembered an important part that I forgot!
I think ...
I vaguely remember something about a convert command that you had to run before doing the upgrade to 6.0 ...

[DavidParker]

Oh, did you follow the Upgrade Document?

This seems pertinent:

12Run the nbpushdata command.
This command moves data from your current database files (a subset of the NetBackup catalog) into a newly created EMM database. The sequence of when and on what systems you run the nbpushdata command is extremely important. For more information on running nbpushdata and creating the EMM database after you upgrade servers, refer to �Populating the NetBackup EMM Database� in the NetBackup 6.0 Installation Guides.

[Dennis_Strom]

I changed convert to recover since I thoght you had it but your are now correct it is the pushdata command. Dang, was working hard to remember that. upgrade to 6 then upgrade to mp3 then pushdata. Plus do not forget to install pbx. Actually you should install pbx before starting the upgrade.

[aske_andersen]

Thank you for the reply.

I just tried:

bprecover -l -dpath path_to_the_5.1_catalog -v

and got:

Created: 11/09/06 20:11:37
Server: xxx

Path
----
IMAGE1 xxx:C:\Program Files\VERITAS\NetBackup\db
IMAGE2 xxx:C:\Program Files\VERITAS\Volmgr\database
IMAGE3 xxx:C:\Program Files\VERITAS\NetBackup\var


Nothing more, do you know the right syntax ?

I would really like to avoide downgrading before upgrading once again
as described in :
http://seer.support.veritas.com/docs/282327.htm



Regards...

[DavidParker]

No problem.
Does this mean that you'd followed the guide through to completion? Including the bppushdata command?

On a side note, if you are satisfied with the support you've received, then please feel free to award some 'Helpful' and 'Correct' points at your discretion.

[DavidParker]

Here are directions on using bprecover:
(Taken from a Unix guide, but the options and flags should be the same on Windows)

bprecover(1M)
NAME
bprecover - Recover selected NetBackup related catalogs.
SYNOPSIS
/usr/openv/netbackup/bin/admincmd/bprecover
-l -m media_id -d density
-l -dpath disk_path
-l -tpath tape_device_path
-l -opath optical_device_path
-r -m media_id -d density

-r -dpath disk_device_path

-r -tpath raw_tape_device_path

-r -opath optical_device_path

Note Stop bpdbm and bprd before using this command. Also, ensure that bpcd the
NetBackup Client service is running on any system that is being recovered.

DESCRIPTION
bprecover initiates the NetBackup utility for restoring the NetBackup internal databases called catalogs and recovers catalogs that were backed up by using the procedures described in the NetBackup system administrator's guide. Use bprecover only if catalogs were destroyed on disk.
The command has two main modes: list and recover. List shows the contents of a backup media or disk path. Recover recovers the catalog files.
You must have root privileges to execute this command.
OPTIONS

-l
Lists the header information from the specified media or disk path.

-m media_id -d density
Specifies the media ID and the density of the media from which to recover files.

ltid and vmd must be running when you use the -m option.
media_id must be six or less characters and must be defined in the Media Manager volume database.

density must be one of the following:
4mm (4-mm cartridge)
8mm (8-mm cartridge)
dlt (dlt cartridge)
dlt2 (dlt cartridge alternate)
qscsi (1/4-in cartridge)
Note The following densities apply only to NetBackup DataCenter servers.
hcart (1/2 Inch cartridge)
hcart2 (1/2 Inch cartridge alternate)
dtf (DTF cartridge)
odiskwm (Optical disk-write many)
odiskwo (Optical disk-write once)

-dpath disk_path

-tpath tape_path

-opath optical_path
Specifies a raw device path. If -m and -d are not specified. you must use

-dpath, -opath, or -tpath to specify a raw device path. Stop the Media
Manager device and volume daemons (ltid and vmd) when using one of these options.

Note Some platforms require a Berkeley-style close device for the tpath option. This is the path with b in the device name (for example on a Solaris system, it could be /dev/rmt/0cbn). You will get an I/O error if you do not specify a Berkeley style close device on platforms that require it.

-r
Recovers images from the specified media or disk path. There are three modes of recovery available with -r:

If -r all (or ALL) is specified, recover all the images contained in the specified media or disk path.
If -r image_number is specified, recover only the selected image number from the specified media or disk path.

If -r is specified by itself, bprecover interactively prompts and asks if you want to recover the images contained in the specified media or disk path.

-stdout
Specifies that the selected backup image be written to stdout instead of automatically being restored. This option is useful, for example, if only one individual file was lost and you want to restore it without restoring the rest of the catalog files in the image.
Note You cannot specify -r ALL with -stdout because the -stdout option permits only one file image to be read at a time.

-dhost destination_host
Specifies the host to which the selected catalog is restored. Normally, catalogs are restored to the host where the data originated (as displayed with the -l option). The -d option makes it possible to restore the catalog to another host.
Caution Use the dhost option with EXTREME caution, since it can overwrite existing catalogs on the destination host. To permit recovery in case you unintentionally overwrite the wrong catalogs, you can move existing catalogs to a temporary directory on the destination host.

The following NetBackup client software must be installed on the destination
host:
/usr/openv/netbackup/bin/bpcd
and
/usr/openv/netbackup/bin/tar
Note Do not specify -r all (or ALL) with -dhost. Either explicitly specify an image
(for example, -r 2) or use the interactive mode (-r).

-v
Selects verbose mode. This is meaningful only when running with debug logging turned on (that is, when the /usr/openv/netbackup/logs/admin directory exists).

EXAMPLES

Example 1
List the backup header information for catalog backup that was done to disk path
/disk/bpbackup.
# bprecover -l -dpath /disk1/bpbackup
Database Backup Information from /disk1/bpbackup
NetBackup Commands
bprecover(1M)
213
Created: 02/20/2002 12:13:47
Server: bphost
Path
----
IMAGE1 /usr/openv/netbackup/db
IMAGE2 /usr/openv/volmgr/database

Example 2
List the backup header information from media ID JBL29, which is density 8mm.
# bprecover -l -m JBL29 -d 8mm
Database Backup Information from JBL29
Created: 01/22/02 07:50:51
Server: bphost
Block size: 32768
Path
----
IMAGE1 /usr/openv/netbackup/db
IMAGE2 /usr/openv/volmgr/database

Example 3
Recover the /usr/openv/netbackup/db files from disk path /disk1/bpbackup.
# bprecover -r 1 -dpath /disk1/bpbackup
Recovering bphost:/usr/openv/netbackup/db

Example 4
Recover all the backed up catalogs from media ID JBL29.
# bprecover -r ALL -m JBL29 -d 8mm
Recovering bphost:/usr/openv/netbackup/db
Recovering bphost:/usr/openv/volmgr/database

Example 5
Interactively restore selected images. Use raw tape path /dev/rmt/1cbn. Assume the
media that is loaded into the drive is the same one as in Example 4.
# bprecover -r -tpath /dev/rmt/1cbn
Recover bphost:/usr/openv/netbackup/db y/n (n)? n
Recover bphost:/usr/openv/volmgr/database y/n (n)? y
Recovering bphost:/usr/openv/volmgr/database

Example 6
Recover a single file from image 1 on JBL29.
bprecover(1M)
214 NetBackup Commands for UNIX
# bprecover -r 1 -m JBL29 -d 8mm -stdout | /bin/tar -xvf
- /usr/openv/netbackup/file_to_recover
Writing bphost:/usr/openv/netbackup/db to stdout

Example 7
Restore an image to another host by using the -dhost destination_host option.
# bprecover -r -m ODL08B -d odiskwm -dhost giskard
Recover bphost:/usr/openv/netbackup/db to host giskard y/n (n)? n
Recover bphost:/usr/openv/volmgr/database to host giskard y/n (n)? y
Recovering bphost:/usr/openv/volmgr/database to host giskard
ERRORS
If any errors occur during the recover operation, error messages are written to stderr.
FILES
/usr/openv/netbackup/logs/admin/*
/usr/openv/netbackup/db/*
/usr/openv/volmgr/database/*

[Dennis_Strom]

before running the recover command try running the nbpushdata command and installing pbx.
http://support.veritas.com/docs/283176

GENERAL ERROR: Running "nbpushdata" on Veritas NetBackup (tm) 6.0 Maintenance Pack 2 (MP2) fails to push the globDB.

[Dennis_Strom]

http://support.veritas.com/docs/283712

GENERAL ERROR: The nbpushdata command can fail if the REQUIRED_INTERFACE entry exists in the vm.conf file on a clustered NetBackup server

[Jason_Voyles]

This information is more of the preventiave steps than fixes, but it might help some people that few this thread from having these problems in the future.
Did you by any chance get to use the NCVU utility before upgrading to 6.0? This utility is very helpful in finding configuration problems that can cause problems during the upgrade.
Also to note, during our upgrades we also ran a catalog consistency check (bpdbm �consistency) before the upgrade to find and resolve any problems with the netbackup catalog and volume database.
I believe both the NCVU utility and the catalog consistency check are suggested to be used by the Netbackup Instillation Guide before an upgrade. We also did due diligence and ran the Netbackup nbsupport script before the upgrade to hand over to support if needed.

As for recovery, is your catalog data in binary or ACSII? It is highly suggested to convert any NetBackup ACSII catalog information to binary mode using the cat_convert utility before upgrading to 6.0. Netbackup 6.0 does not support writing to ACSII catalog data. It is however noted that it should be able to read the data, however I do not know many people who have ACSII catalog data configured.
If you upgraded the box from 3.4 or 4.5 to 5.1, then this might be an issue. Default behavior on 5.1 is to use Binary catalog data.

Do you see the directories of your clients in "../NetBackup/db/images" ? Are there directories and files in the "images///" folders?
The files will be in the format of "__" and you should see many that end in ".f"
If these files are present, then your catalog data is intact and there is a software problem. If the data is missing, then your catalog has been wiped during the upgrade of netbackup somehow and will need to be recovered from tape.

I wish you much luck.

[aske_andersen]

Jason,
We didn't use the NCVU, I've not been able to locate it on my HD !?!
My Catalog must be in Binary, as we made a clean 5.1 install 1.5 years ago.

In
/NetBackup/db/images, we see all the folders i.e.
a folder called Exchange2003 this folder contains:
other foldes(probably the date stamp you referred to) called:
1142000000
1157000000 etc. etc.

And those folders contains files in __ format
(as you wrote) - some of them ending with .f

I other words the Catalog should be intact - what will be the next step ?

I've been running the bpdbm -consistency
It complains about indexlevel missing (for all the servers)
i.e:

check directory of client
________________________________________________
INDEXLEVEL for exchange2003 was not found
checking image file
checking files file

The Indexlevel missing is properly due the fact that our catalog are missing ?

[DavidParker]

Aske,
Here are directions for how to download (it's doesn't come with NBU by default) and use NVCU:

http://seer.entsupport.symantec.com/docs/244019.htm

[Jason_Voyles]

The errors during a catalog consistency check for "INDEXLEVEL for was not found" is an expected error if you're not doing catalog indexing. Catalog indexing is not enabled by default and must be manually configured by creating some touch files in the catalog path. Unless you're experiencing problems with catalog performance, indexing is generally not needed.
Also to note, There would be catalog information in the directories I mentioned above if you had a successful backup. Check the time stamps on the files to see when they where created. If you have files that are there previous to the upgrade, then the catalog data is there, if not then you need to recover the catalog data from tape.