Restore from more masters

[Michael_STRAND]

Hi everybody, after days of struggling I decided to ask here, as I believe there has to be some very simple solution which is eluding me.

My Linux client has multiple SERVER entries in bp.conf for our multiple master servers. I'm trying to accomplish restore of Oracle DB from different client through rman. When I request

ALLOCATE AUXILIARY CHANNEL c1 DEVICE TYPE 'SBT_TAPE' parms="ENV=(NB_ORA_CLIENT=<original_client>,NB_ORA_SERV=<master_having_original_client>)";

I can easily restore db with duplicate command but only as long as NB_ORA_SERV is first in bp.conf. Whenever I need to restore from different master, my backup sets are not found. So, to make it simple:

How can I choose from multiple master servers which catalog is going to be used, if not with NB_ORA_SERV?

[Nicolai]

According to this TN setting NB_ORA_SERV is the right thing to do in yore case:

Source: http://www.symantec.com/docs/TECH58933

NB_ORA_SERV
This is an optional parameter and will default to the first SERVER entry in the bp.conf file.  Set this parameter only if this backup or restore must use a master server that is different from the one used for standard backup and restore operations on this host.

Do you have local bp.conf files in /oracls/SID/???

update:

Check also "About the order of precedence for runtime configuration variable
settings when using scheduled backups with templates". This describes the order of precedence for the runtime configuration variables

http://www.symantec.com/docs/DOC5175 Page 57

By precedence means that if the environmental variable $NB_ORA_SERV is set, it will have precedence over NB_ORA_SERV set in the RMAN send section. This could easily explain the behavior you see.

[Michael_STRAND]

Hello Nicolai, thanks for helping me out. Exactly as you said and documentation confirms, NB_ORA_SERV is the parameter which should do the trick. Now its pretty weird it doessn't work. Even weirder is that NB_ORA_CLIENT is being accepted, otherwise no restore would work. What you mean with "bp.conf files in /oracls/SID/???"

I also verified that $NB_ORA_SERV is not set.

[Michael_STRAND]

Now it all went south. I am not able to restore anything anymore from my second master to my test server, even though I put only one master server in bp.conf and restarted netbackup service. I am not aware of any other change. Client is authorized, has policy assigned and accessible from master server. This is killing me, I started two weeks ago and I didn't move a bit since.

[Marianne]

What exactly is restore failing with?

Do you have bprd log folder on the master?
You need bprd log to see if restore request is received from client and how master is interpreting the request.

Is master also the only media server?
If not, you need SERVER entries for all media servers that are allowed to backup/restore the client.

[Michael_STRAND]

My wrong, forgot to share the message:

channel c1: starting datafile backup set restore
channel c1: restoring control file
channel c1: reading from backup piece c-3101352011-20140522-00
channel c1: ORA-19870: error while restoring backup piece c-3101352011-20140522-00
ORA-19507: failed to retrieve sequential file, handle="c-3101352011-20140522-00", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   Failed to process backup file <c-3101352011-20140522-00>

Let me recheck the whole config and the logs.

[Marianne]

Does this restore attempt make it to Activity Monitor? In addition to bprd on the master, ensure that dbclient log folder exists on the client. If Unix/Linix, remember 777 permissions.

[Michael_STRAND]

I"ll recap whole situation. I have two masters, nb1 and nb2. My client test1 was always connected to nb1, where everything works as expected. It was possible to redirect it to nb2 with stopping netbackup service, changing SERVER = nb1 to SERVER = nb2 and starting service again. nb2 has policy with test1 assigned, so everything worked right away. While I tried to resolve the original issue in the first post, I introduced new error. After redirecting with just described way I can ping test1 from nb2 console's Client list, but rman restore fails with this:

channel c1: ORA-19870: error while restoring backup piece c-3101352011-20140522-00
ORA-19507: failed to retrieve sequential file, handle="c-3101352011-20140522-00", parms=""
ORA-27029: skgfrtrv: sbtrestore returned error
ORA-19511: Error received from media manager layer, error text:
   Failed to process backup file <c-3101352011-20140522-00>

Nothing appears in nb2 Activity monitor. nb1 still works as always.

edit: Both nb1 and nb2 are both master AND media servers at the same time.

[Michael_STRAND · Accepted solution]

Finally, got it both. One of masters had incorrect altnames and the other had a problem with vlan. This was a tricky one, but it finally works as expected.

[Nicolai]

Thanks Michael for providing the resolution to the issue.