cancel
Showing results for 
Search instead for 
Did you mean: 

Restore Netbackup catalog to a test server?

SteveLaw
Level 5

We're preparing to upgrade our Netbackup master and media servers from v7.5.0.6 to 7.6.1.2. Part of the process involves upgrading the catalog and according to the documentation this process could take anywhere from 6 - 14 hours. Our master is a VM and the disk performance doesn't seem to be very good. I'd like to run through the whole upgrade process and get some idea of how long the catalog conversion will take when we upgrade the live system. There are settings that can be applied that are supposed to speed up the process etc.

I cloned the master in the hope that I could run through the process on the clone, but for unclear reasons this doesn't work - the cloned master has to be on an isolated network, and it's a domain server that can no longer reach the domain, and this makes Netbackup unhappy - it complains that the clone is not the original server, and it refuses to run or services continually shut down due to errors.

Therefore my only option is to create a new temporary master (with a different server name) and restore the catalog from the live master to this test server. Is this feasible?

We use Data Domain disk storage in the live system, but I don't see how we could restore to the new test master from Data Domain. I'm thinking a I could create a basic disk on one of the media servers and write the catalog to that, then switch that media server to the new master and read the catalog that way. But I would also need the small DR file that the catalog backup creates, and surely the new master won't be able to process that, as it was written by the Netbackup master in our live environment?

Am I going about this all wrong? Any thoughts and suggestions gratefully received :) 

 

1 ACCEPTED SOLUTION

Accepted Solutions

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

What about using a CIFS share? 

 

View solution in original post

12 REPLIES 12

mph999
Level 6
Employee Accredited

You cannot restore a catalog to a server with a different hostname.  Unfortunately, the server name is deeply embedded in the NBDB.

The rules for catalog restore:

1. NBU must be installed in exactly the same path

2. NetBackup version must be exactly the same

3. Hostname must be identicle

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

I have suggested in your previous post over here that you add an entry in the test server's hosts file.

Have you tried that?

Something like :
<IP-address> swn009.corpds.net  swn009 

Then install NBU specifying wn009.corpds.net as master and EMM server.
(See rest of requirements in Martin's post.)

Rather add a Basic Disk (USB or CIFS share) to the master server for the catalog backup which you can then attach to the test master (using exact same path/mount point as original server).
Backup to a media server will add additional complexity when trying to restore catalogs to test server. 

 

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Another 2c from my side:

PLEASE skip NBU 7.6 and go straight to latest 7.7.x.

ALL NBU versions up to 7.6.x will reach EOSL in less than 6 months. 
See https://sort.veritas.com/eosl/show_matrix 

 

Hi Marianne,

I did take note of your response on my previous post ( wow, you keep track of everything!).
We don't actually have an isolated test network in vmware. I first brought the cloned swn009 master up with the vm nic disabled then added the hosts entry, but Netbackup still refused to start. It seemed that with the nic disabled it wasn't reading the hosts file. So I then enabled the nic and changed the ip address to something else with no gateway, so it couldn't route, then I added that a hosts file entry so the new "fake" ip resolved to the master hostname. That seemed to work, but I found that if Netbackup was left running for a while services would stop and errors came up and if I restarted Netbackup most of the services failed to start.
Maybe I just needed to cancel all jobs, disable all SLPs etc....? 

You wrote:
"Something like :
<IP-address> swn009.corpds.net  swn009 

Then install NBU specifying wn009.corpds.net as master and EMM server.
(See rest of requirements in Martin's post.)"

Are you saying that I can install Netbackup to a different host but trick Netbackup into thinking that server has the same hostname as the original master by adding a hosts file entry? e.g. if new test server is SWN011 IP 10.0.0.2 and original master is SWN009 IP 10.0.0.1, then I add a hosts entry on SWN011 that says "10.0.0.2  SWN009", then install Netbackup? 

 


Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

I would not personally try that.

I would only install to hostname swn009  with different IP address and network enabled and FQDN as alias in the hosts file.

After installation, bpclntcmd -self must list swn009.FQDN linked to swn009 IP address.

Please note that NBU is also case sensitive.

SWN009 != swn009 != swn009.corpds.net  

Whatever the NBU master server hostname is on the production server, must match and be in place before you start with installation.

As soon as NBU is up and running, deactivate all media servers, policies and SLPs. 

 

 

Hi, sorry to be such a poor correspondent! So freaking busy...

If I build an isolated replica of the master server, how do I restore the catalog to it? Would I:

*Create a basic disk for backing up to on the master (will probably need to temporarily add a new drive as the catalog backup is 500gb). 
*Run a catalog backup to the new drive
*Create a matching new drive on the isolated duplicate of the master
*Copy the catalog backup from the live master to the duplicate somehow (probably create an iso?)
*Use the catalog dr file from the live master to run the catalog restore on the duplicate master.

Does that sound like a plan?  

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Can you not get hold of a USB drive?

Create a Disk Storage Unit for the USB drive and write catalog backup and DR-file to this drive.

Map the USB drive to same drive letter on test server for catalog recovery.

 

Hi,

If you have a gap in your backup window of 6 hours, the best and most accurate way that I can see you performing this 'Test Upgrade' would be to take a snapshot of your current VM master server after stopping SLP's and prevent any backup jobs from processing (see below)

To prevent user initiated jobs from starting:
/usr/openv/netbackup/bin/admincmd/bprdreq -terminate
To prevent scheduled backups from starting:
/usr/openv/netbackup/bin/admincmd/nbpemreq -suspend_scheduling
 
To resume processing of both user initiated and scheduled jobs:
/usr/openv/netbackup/bin/initbprd
/usr/openv/netbackup/bin/admincmd/nbpemreq -resume_scheduling

 

Perform the upgrade to the latest version of NBU (as per supported upgrade path documentation).  If this is succesfull you can then re-enable the SLP's and job processing and no further work is required. 

Remove the snapshot after a period of time from vCenter (I would recommend 1 month but that is me).

Cheers,

J

Hi Marianne, these are virtual machines, so can't just hook them up with a usb drive. 

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

What about using a CIFS share? 

 

Hi J,

I'd probably clone it rather than snapshot, just to be sure there would be no issues returning to the pre-snapshot phase. 

I have tried this already but as per original post I had problems tricking Netbackup into thinking this was the original server (once cloned it could not connect to the domain and had no ip connectivity) and Netbackup services kept shutting down. Maybe the trick is to suspend all processing as per your commands. 

Hi,

If you are using ESX, then a snapshot if you have the capacity and the open backup window.  This has to be the entire virtual machine, you cannot snapshot the machine without the MSPD volumes.

When using a clone, you will have a new/different server, your IP address, server name, machine UID etc are all different.

Using a snapshot for the entire machine will allow you to return to the exact point in time from when the services are stopped, pre upgrade in the event of an upgrade failure.

If the upgrade is successful you only need to remove the snapshot after a suitable period of time.

If the upgrade is unsuccessful then you may require assistance from Veritas Support to perform your upgrade. (depending on the errors).

Cheers,

J