NDMP Directed Recovery

[sunixadm]

Hi,

We have decided to use a dedicated EMC² Celerra Data Mover for our NDMP recoveries and after reading the NetWorker documentation, there are questions to which I can't find (or don't understand) the answers to.

We are using NetWorker 7.1.1 and the server is a Sun E450 running Solaris 8.

We have a production CIFS Data Mover named server_1. server_1 sees and uses our tapes via an FC connection and DDS.

We have a dedicated Data Mover for recoveries named server_2. Server_2 sees and uses our tapes via an FC connection and DDS.

Both Data Movers have dedicated sotrage from a Symmetrix.

We are backingup the data from server_1 using NDMP and wish to continue doing so.

Because of the CPU load caused during a restore, we would like to use server_2 to perform these recoveries.

The questions are:
1. When I perform a directed recovery using:
nsrndmp_recover -c server_1 -s backupserver -R server_2
how do I specify what I want to recover and how do I specify the path to recover to on server_2? The -r and -m switches seem to be the right candidates but, when I read the definitions of these switches, I'm not really sure.

2. Will server_2 use its resources (CPU and memory) for the tape operations? This is the entire point of this exercise! This isn't adressed in any of the NetWorker documentation. I have asked this question of EMC² support but, they turned me to Legato. Legato is still looking into this but, I thought I might get a quicker response from someone here.

Thanks!!

-Joe

 

[605]

1. Just append the pathname(s) to the command (and use the option -F). The manpages should tell you more details.

2. I do not think so. The idea about NDMP data transports is that such 'devices' can move data directly without any controlling instance involved. It is just providing an NDMP tape interface for a SCSI tape drive.

 

[michaelw70]

does the -R indicate a directed recover? We do ndmp recoveries and put them in different locations on our celerra filesytem using the -m command for saveset recoveries from the command line. The gui works too if you have the patch that fixes a bug they were having.

My command looks like this:

nsrndmp_recover -c client -s servername -S ssid -m celerra1::/restore /fs1/filetobe/restored

new location comes first and files to get restored come second. When talking with the legato tech, she told me if i wanted it to go to an entirely different server just change celerra1 to the appropriate server. If any spaces are in the names or folders you have to quote it.

question for you?? By doing it this way, do you require 2 NDMP licenses?? Our scenario, we use fs_timefinder and mirror the filesystem to BCVs on a second datamover, and then backup the mirror and do restores to the mirror.