Restore after Media Servers Decommed

[marcymarc24]

NB 6.0MP6 on Solaris 10 master servers.

We have a number of media servers unavailable after a site move.  We would like to restore data from tapes associated from these servers, I am guessing they weren't decommissioned 'cleanly'.

Is it a case that the tapes I need to restore from have to be imported and I also need to add a redirect into the conf file?

-Marc

[Max_Booth · Accepted solution]

You don't need to import the tapes since the images will still be in the catalog and the tapes in EMM, you should just be able to set the Media Override host via general settings in the GUI or the setting in bp.conf on the master: FORCE_RESTORE_MEDIA_SERVER <oldserver> <newserver>

You should look to get this cleaned up though, but that should do you for the time being.

Regards,

max.

[Andy_Welburn]

DOCUMENTATION: What is the process for decommissioning a NetBackup 6.0 media server?
http://www.symantec.com/business/support/index?page=content&id=TECH52021

Possibly a bit "stable door & horse" I know but....

Also, step 4 of the above (as stated by Max):

"...
Use the Administration Console to configure NetBackup so that restore requests are directed to <new_server>.  In the General Server host properties of the master server, add a Media Host Override entry that specifies <old_server> for the Original backup server and <new_server> as the Restore server.
..."

[marcymarc24]

I've added the redirects in the bp.conf (and restart the software).  I've commented out references to the old media servers also.

Despite this the restore still hangs, I guess it is trying still to communicate with the old media servers?  Below is an extract from the bprd log.  Any ideas on what else I could try?

09:47:06.681 [29844] <2> vnet_async_connect: vnet_vnetd.c.4163: getsockopt SO_ERROR returned: 145 0x00000091
09:48:04.057 [29844] <2> vnet_async_connect: vnet_vnetd.c.3752: ran out of time before connect: 302 0x0000012e
09:48:04.057 [29844] <2> vnet_async_connect: vnet_vnetd.c.3752: ran out of time before connect: 302 0x0000012e
09:48:04.058 [29844] <2> vnet_connect_to_service_or_vnetd: vnet_vnetd.c.3521: vnet_async_connect failed: 18 0x00000012
09:48:04.058 [29844] <2> nb_connect_to_vnetd_or_legacy: comm.c.1971: vnet_connect_to_vnetd_or_service failed: 18
09:48:04.058 [29844] <2> bpcr_connect: nb_connect_to_vnetd_service(xxxxxxnbumed04) failed: 25
09:48:04.058 [29844] <2> bpcr_connect: Can't connect to client xxxxxxnbumed04 09:48:04.058 [29844] <2> ConnectToBPCD: bpcd_connect_and_verify(xxxxxxnbumed04, xxxxxxnbumed04) failed: 25
09:48:04.058 [29844] <2> start_mpx_group: Can't connect to xxxxxxnbumed04. Errno  = 150: Operation now in progress
 

[Andy_Welburn]

Is "xxxxxxnbumed04" the 'decommissioned' media server?

[marcymarc24]

Yes it is, this is commented out in the conf file.

-Marc

[Andy_Welburn]

so I hope not to lead you astray, but I would remove the decomm'ed media totally from bp.conf as opposed to just commenting out (in all honesty I'm not sure how remarks are interpreted here.....?)

Is the syntax for your force restore entry correct? Maybe try it via the GUI?

Hoping for a few more comments from those that have had experience here...!

[marcymarc24]

Syntax used for the redirects in the conf file below...

FORCE_RESTORE_MEDIA_SERVER xxxxxxnbumed01 xxxxxxnbumst
FORCE_RESTORE_MEDIA_SERVER xxxxxxnbumed02 xxxxxxnbumst
FORCE_RESTORE_MEDIA_SERVER xxxxxxnbumed03 xxxxxxnbumst
FORCE_RESTORE_MEDIA_SERVER xxxxxxnbumed04 xxxxxxnbumst
FORCE_RESTORE_MEDIA_SERVER xxxxxxnbumed05 xxxxxxnbumst

-Marc
 

[Andy_Welburn]

except, I would go through the decomm procedure & remove all references to any decomm'ed media servers.

The other advantage of this is that performance of such things as the GUI can be affected if they remain - e.g. NB tries to get info from media servers when opening the GUI or expanding host properties etc and there can be severe delays if it tries to connect to a media server that no longer exists.

[Marianne]

bp.conf entry should look like this:

FORCE_RESTORE_MEDIA_SERVER = fromhost tohost

Please ensure you have bprd log on the master server. (Restart NBU after creating log dir).

Look for the following lines in bprd log and verify that your interpretation of what needs to be done is the same as NetBackup's:

.... <2> restorefiles:    browse_client =
.... <2> restorefiles:    requesting_client =
.... <2> restorefiles:    destination_client =
.... <2> restorefiles:    requesting_client_hostname =
.... <2> restorefiles:    destination_client_hostname =
.... <2> restorefiles:    requesting_user =
.... <2> restorefiles:    requesting_group =
.... <2> restorefiles:    progress_file =
.... <2> restorefiles:    rename_file =
.... <2> restorefiles:    policy =
.... <2> restorefiles:    client_type =

Also see in bprd log if FORCE_RESTORE... entry is found and interpreted correctly.

[Andy_Welburn]

that's what you get from not reading things through carefully!

At least it wasn't a [SPACE] that was missing or extra!

[marcymarc24]

It was either the missing '=' that resolved this or removing the old media servers no longer in use from the conf file.

In any case many thanks.

-Marc

[Andy_Welburn]

that did the trick, glad you managed to get it working!

Hard to say who should get the solution for that:

Max for suggesting the entry in the first place  (( )) ;

Me for saying check the syntax of same ; or

Marianne for actually noticing the syntax was incorrect!

I'll let you guys fight that one out!

[Marianne]

My vote is for Max who suggested the FORCE_RESTORE entry.

If you look at bprd log, you will see that there's an entry that says something like this: "because of FORCE_RESTORE.... changing media server to yyyyyy"

**** Would be nice if we could all share the points, though...... ****

[RiaanBadenhorst]

You both got a whole bunch of thumbs up just now, hope that helps :)