Unsolved

This post is more than 5 years old

22 Posts

11355

March 18th, 2007 09:00

Failed to load SBT library {ORACLE_HOME}/lib/libobk.so

hi,

We have oracle 10g RAC on 2-nodes solaris x86-64 bit, we installed networker storage node and networker module for Oracle on each node.


We have been throught the steps for rman backup and customised the nsrnmo file.
we have linked the library as oracle user:
ln -s /lib/libnwora64.so libobk.so
please note we choose to install the S/W on the default path, but the path for libnwora64 was /lib/ and not /usr/lib

we run the following commands on the rman
RUN
{
connect target rman/rman@zafad
ALLOCATE CHANNEL c1 DEVICE TYPE sbt
PARMS='SBT_LIBRARY=/opt/app/oracle/product/10.2.0/db_1/lib/libobk.so ENV=(NSR_SERVER=svr-bkp,NSR_GROUP=Oracle)';
BACKUP CURRENT CONTROLFILE;
}
and the output was
RMAN-00571: ===========================================================
RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============
RMAN-00571: ===========================================================
RMAN-03009: failure of allocate command on c1 channel at 03/18/2007 19:23:55
ORA-19554: error allocating device, device type: SBT_TAPE, device name:
ORA-27211: Failed to load Media Management Library
Additional information: 25


Please advice

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 18th, 2007 16:00

It's library. First make sure you have good NMO release as there are 3 different for Linux (same goes for client). I recently had RAC installation on EM64T and had no issues. SBT library should not be defined in RMAN script, but in nsrnmo. Also I believe Oracle prefers using send instead of parms now.

2 Intern

 • 

2K Posts

March 18th, 2007 22:00

Recheck libbrary linking and remove SBT_LIBRARY from script.

Run this test script and it should show NMO or MML based on NMO version you are using to signify that linking has been done proerpy. Then test backup with the backup script.

Test script -

connect target user/password@{database};
connect rcvcat user/password@{recovery catalog};
run {
set command id to 'abc';
allocate channel t1 type 'SBT_TAPE'
parms 'ENV=(NSR_SERVER={backup server},NSR_DATA_VOLUME_POOL=Default)' ;
allocate channel t2 type 'SBT_TAPE'
parms 'ENV=(NSR_SERVER={backup server},NSR_DATA_VOLUME_POOL=Default)' ;
release channel t1;
release channel t2;
}

Backup Script -

connect target user/password@{database to be backedup};
connect rcvcat user/password@{recovery catalog};

run {
change archivelog all crosscheck;
allocate channel t0 type 'SBT_TAPE' parms 'ENV=(NSR_DATA_VOLUME_POOL=Poolname)';
allocate channel t1 type 'SBT_TAPE' parms 'ENV=(NSR_DATA_VOLUME_POOL=Poolname)';
allocate channel t2 type 'SBT_TAPE' parms 'ENV=(NSR_DATA_VOLUME_POOL=Poolname)';
allocate channel t3 type 'SBT_TAPE' parms 'ENV=(NSR_DATA_VOLUME_POOL=Poolname)';
backup full filesperset 4
format '/FULL_%d_%u/'
database
include current controlfile
;
sql 'alter system switch logfile'
;
backup
(archivelog all);

release channel t0;
release channel t1;
release channel t2;
release channel t3;

}

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 19th, 2007 07:00

Of course, because you still have the same issue - script won't change that. Can you post uname -a from both boxes? Also, are you sure you installed appropriate NMO on both RAC machines?

22 Posts

March 19th, 2007 07:00

Hi,

I run the srcipt and i got the following error:

Recovery Manager: Release 10.2.0.1.0 - Production on Mon Mar 19 17:05:59 2007
Copyright (c) 1982, 2005, Oracle. All rights reserved.
RMAN> connect target rman/rman@XXXX;
connected to target database: XXXX (DBID=223369585)
RMAN> connect rcvcat rman/rman@YYYY;
connected to recovery catalog database
RMAN> run {
2> set command id to 'abc';
3> allocate channel t1 type 'SBT_TAPE'
4> parms 'ENV=(NSR_SERVER=svr-bkp,NSR_DATA_VOLUME_POOL=Default)' ;
5> allocate channel t2 type 'SBT_TAPE'
6> parms 'ENV=(NSR_SERVER=svr-bkp,NSR_DATA_VOLUME_POOL=Default)' ;
7> release channel t1;
8> release channel t2;
9> }



executing command: SET COMMAND ID
starting full resync of recovery catalog
full resync complete


RMAN-00571: ===========================================================

RMAN-00569: =============== ERROR MESSAGE STACK FOLLOWS ===============

RMAN-00571: ===========================================================

RMAN-03009: failure of allocate command on t1 channel at 03/19/2007 17:07:28

ORA-19554: error allocating device, device type: SBT_TAPE, device name:

ORA-27211: Failed to load Media Management Library

Additional information: 25

22 Posts

March 19th, 2007 08:00

Hi,

uname -a for RAC1:

SunOS hostname-rac1 5.10 Generic_118855-33 i86pc i386 i86pc

uname -a for RAC2:

SunOS hostname-rac2 5.10 Generic_118855-36 i86pc i386 i86pc


I installed on both RAC the same module (nmo42_solaris_64)

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 19th, 2007 13:00

Silly me, I just realized it is on x86 Solaris (somehow I skip that part and thought your were on Linux).

By saying that I'm also confused by where did you get NMO for Solaris x86. There are 32bit and 64bit versions, but they are for SPARC and not x86 AFAIK. Did someone tell you to use SPARC version?

2 Intern

 • 

2K Posts

March 19th, 2007 19:00

Apart from the correct NMO, also recheck on the library linking.

May be you can use sbttest for that.

22 Posts

March 20th, 2007 03:00

Hi,
I thought that NMO for Solaris 32bit and 64bit works on SPARC and on X86, no one informed me that the module works only on SPARC.

Do you have any idea which module i have to use for the Solaris X86???????

Thanks

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 20th, 2007 03:00

None I'm afraid, but you better check that with support. NMO 5 should be in beta and perhaps it supports Solaris x86.

37 Posts

March 21st, 2007 12:00

Hello,

I am a colleage of Johnny, and as you advised guys we contacted support, and it seems that the NMO 4.2 does not support Solaris x86 64-bit version, maybe the newer version will do so.

But we were advised to try the NMO 3.5 for Solaris x86 (32-bit), we were told it might work on Solaris x86 (64-bit).

The problem now is that I couldn't find the module itself, any ideas?/?? please advise.

Thank you
Ziad

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 21st, 2007 14:00

Did support gave you that advice? 3.5/3.6 did indeed support Solaris x86 and version is hidden on ftp.legato.com - if support gave you this advice they should also give you the link. Did they say they would support it too?

2 Intern

 • 

131 Posts

March 22nd, 2007 02:00

Hi,

we have installed the NMO 3.5 and did a relinking.
before nsrnmo was not even starting, but now the nsrnmo is starting but in a certain stage it gives the following error:

Segmentation Fault - core dumped
nsrnmostart returned status of 139
/usr/sbin/nsrnmo exiting.

any suggesstions.

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 22nd, 2007 03:00

To be honest, I can't see how NMO3.5 would be able to cope with Oracle 10. That's why I asked if support would support it ;)

2 Intern

 • 

131 Posts

March 22nd, 2007 04:00

Hi,

well after installing the NMO 3.5 and trying the backup, it is saving the indexes but the following error is appearing:

3 retries attempted
No full backups of this save set were found in the media database; performing a full backup
RMAN> 2> 3> 4> 5> 6> 7> 8> 9> 10> 11> 03/22/07 14:50:12 NetWorker: Error: process 23737 running command "/opt/app/oracle/product/10.2.0/db_1/bin/rman" completed with an error code of 1.
nsrnmostart returned status of 1
/usr/sbin/nsrnmo exiting.

any progreesion you see, and if yes any suggesstions.

6 Operator

 • 

14.4K Posts

 • 

56.2K Points

March 22nd, 2007 05:00

No progression. Index gets saved each time after backup - successful or not (unless you change it).
No Events found!

Top