cancel
Showing results for 
Search instead for 
Did you mean: 

Disaster recover with PureDisk 6.6 and NBU 7

nlc_ICT
Level 2
Hi All,

I've been looking around forums for a while now but no one seems to eb doing anything similar to what I'm after. I need to do a mock disaster recovery of our NBU master server from the backups on an off-site pure disk server. I did not do the original configuration of the process and know next to nothing about puredisk and NetBackup, so I've been kind of thrown in the deep end here.

I'll describe the topology of our backup architecture. We have 1 media/master server (Win 2k8 r2 x64, running NBU7) the NBU master/media server is connected to a local Puredisk diskpool (SPA). The Master server stores backups on the local SPA. This SPA is then replicated to an off-site Puredisk SPA with a PDDO job. Backup catalog files are copied to the off-site Puredisk  server using a third party SCP client.

For the purposes of this mock up, we are assuming that the whole building has burned down and are only left with the offsite backup. to simulate this I have configured a VM running the same OS and disk partition configuration as the NBU master server. the VM is not joined to the domain (unlike the current Master server), and has the same computername as the master server. I have edited the hosts file so that the the FQDN of the master server refernces its own IP, and the IP address of the off-site Puredisk server referes to the FQDN of the onosite Puredisk server (in an attempt to make NBU think they are the same server... not sure if this is correct or not).

 I have then installed NBU7 in the location that it was installed on the actual master server, configured the diskpool to link to the off-site SPA (using the FQDN of the onsite), and attempted to recover the catalog files on the new master server locatedo n the VM. However, this last step continually fails.

Anyone able to tell me what I'm doing wrong?
1 REPLY 1

AAlmroth
Level 6
Partner Accredited
You should be able to recover the NBU catalog using the DR wizard (or bprecover command). As long as you transfer the online catalog backup to the DR site, the restore should work fine. The main issue is rather that there could be a problem with different IP addresses. If PD lists IP addresses rather than hostnames in the GUI, you would need to use the same IP address on the DR site.

The re-occuring issue with PDDO is that when you register the NBU server with the PD SPA, you end up with different disk references. You may see @aaaab on the primary site for the same site PD, and @aaaag for the DR site. But when you do the same on the DR site master server you may end up with @aaaaf for the DR site PD SPA. Now, the catalog is referencing image on @aaaab (or @aaaag) but the DR site master does not now this reference, and hence restores fail. This scenario happens when you use some copy tool to delta copy the image catalog (thus not an online catalog backup). Also, a partial restore of the catalog (when EMM is not included) would end up with the same failure.

So, if you dont recover the catalog, and use two separate NBU masters, your only option would be to do a two-phase import of all images from the DR site PD to the DR site master. You can run this as a cron job or Scheduled Task for the phase #1 import. Then upon DR, do the phase #2 for relevant systems.

There is work-in-progress to automate this using PD based catalog backups, but we will just have to wait I guess. Doubt it will make it to 7.0.1 though.

/A