We have two Networker installations, one in Europe and one in the US. We have NDMP DSA backups to tape made in the Europe location that we want to import and recover to the US location.
I know that using scanner to import indexes is not possible with NDMP backups, but what is the best way to proceed? Do we need to create a new data mover on our Celerra that matches the old host name?
Another wrinkle is that the tape media IDs from the Europe location are also in use in the US location (e.g. 000010L4) - so how do I get rid of the current volume information and import the new ones without wrecking the old ones?
Please let me know if I can clarify - thanks in advance!
You can recover NDMP backups from a Celerra to a different Celerra with a different hostname. You need to specify the hostname AND directory when you specify the relocation directory.
If you know the path of the file or directory that you want to recover, then you can use saveset recover to recover some or all of the data. Otherwise, without the client file index for that data, you will have to recover the entire backup. Note that the file and directory names are case sensitive.
The volume conflict is your biggest problem. The volumes with the same volume name will almost certainly have different volume id. So if you scan the tape, NetWorker will record the volume name as ~volume so that it does not conflict with the volume already in the media database. This name not be a bad thing since you simply want to register the tape into the media database and recover the data.
Before you scan the tape in using scanner -mv, make sure you have a backup of your bootstrap. This will save your media database before starting. Then after scanning the tape and recovering the data, or if something does go wrong, you can recover the bootstrap, and you're back to the point before scanning the tape.
Note: Please test the following by backing up and recovering a small amount of files. This will verify that the information is still current and accurate.
Note: esg70854 referes to nwadmin, which is no longer used since NetWorker 7.3 was released, and was replaced with NetWorker Management Console.
To perform a saveset recover of NDMP backups from one filer (such as a Celerra) to another filer of the same type:
Using command line:
(extracted from "How to use nsrndmp_recover command to implement Saveset recovery" esg91815)
Client file indexes are removed after they pass their retention date. The 'scanner -i' command with NDMP will not rebuild client file information. Therefore, use the scanner command with -m option to generates media database entries. Once the NDMP savesets are added to the media database, you can perform a Saveset Recover to restore NDMP data with following command
nsrndmp_recover -s <networker_server_name> -c <client_name> -S <ssid> -v off -m <hostname::/path_to_recover_to> <path_to be_recovered>
Where: <networker_server_name> is the hostname of the NetWorker server,
<client_name> is the client hostname (NDMP server: filer/datamover ) that this data originated from
-m option uses for new destination either on the same NDMP server but different directory, or the different NDMP server
<hostname::/path_to_recover_to> hostname of the NDMP server (filer/datamover) you are recovering to followed by 2 colons and
the full path starting with the mountpoint down to the directory you would like to recover this data to
The directory in path-to-recover-to MUST be existence on the host.
<path_to_be_recovered> is the full path name starting at the mountpoint down to the directory you would like to restore.
Using the GUI:
(extracted from "Goal: Perform a directed NDMP recover" esg70854)
Fix: 1. Rules for directed recoveries: (A) The O/S used for the backup must be the same at recovery time. That is, if you backed up to a EMC Celerra, you must recover to the same or another Celerra. (B) The target machine to recover to must be defined as a NetWorker client (set NDMP to Yes, specify the remote access, remote user and password fields) (C) Ensure that the remote user defined for the target client has permissions to write to the specified directory(ies) (D) Use one the methods described below to recover and relocate data
Fix: I.Performing a directed index recovery via nwrecover (UNIX):
A. Select the client from which you are recovering (Change->Client)
B. Select the file(s) you wish to recover
C. Select 'Options'
D. In the 'Relocate to:' field, type in <target_hostname>::/<mount_point to restore to>
For example, if you wished to relocate a file from client 'NAS1' to a location named '/fs1/mydir' on 'NAS2', you would type in: NAS2::/fs1/mydir
E. Respond to the other prompts as you would for a regular recovery and click 'Start'
Fix: II. Performing a directed saveset recover from the UNIX 'nwadmin' GUI:
a. From the main menu click on 'Save set' and then 'Recover'.
b. Select the client and then instance to recover from
c. Click 'Recover'
d. Specify the path to relocate (if selecting files or folders from this saveset)
e. To relocate the recovered files, click on 'Options' (else click 'Start' to recover the files to their original location).
f. If you wish to relocate to another location on the same filer:
- Do not change the destination client name
- Specify the full path you want to relocate to. It must of the form /vol/...
- Copy the same exact path to the 'raw device' field
g. If you wish to relocate to another filer (of the same type):
- Make sure you have a client resource for the second filer in the NW server
- In the options window, type in the name of the filer to recover to as the 'Destination Client'
- Specify the path on the second filer to relocate to. Make sure it exists, is accessible and is of the form /vol/......
- Leave the 'raw device' field blank
- Click OK and then 'Start' in the previous window
h. Upon a successful restore, you will find the file(s) relocated to the specified location on the target machine
Fix: Note: If the original filer that the data was backed up to is currently unavailable, but is still listed as the storage node in the client resource, then at the time of the recover, it will need to be changed to the name of the new filer to which the ndmp device is now attached (if data was backed up using an ndmp device).
Hi Wallace - thanks very much for the information! That has been very helpful. I ended up deleting the old volumes with nsrmm and am in the process of scanning in the new volumes.
I have one more question about nsrndmp_recover: can the destination directory that I recover to on the destination host be a directory that I create in Windows?
For example, our Celerra NAS export is /root_vdm_x/export on the destination host. I created a directory there in Windows with the path \\export\restore. So would it work to use /root_vdm_x/export/restore as the destination directory for nsrndmp_recover?
Thanks again! - Greg
In your example /root_vdm_x/export is the directory on the destination Celerra host. The fact that it is mapped out to a Windows host is not relevant for this recovery (at least I don't think so), because the recovery is actually done on the Celerra itself.
Make sure you create the NetWorker client for the Celerra on the server you are scanning into.
Hope the tape you are scanning in is part of the same pool that the "other" tape with the same volume name is in.
Again, do some testing... backing up some trivial directory, then recover with relocate to a different area. Works fine in the docs, but docs won't help if it goes wrong.