cancel
Showing results for 
Search instead for 
Did you mean: 

Backup node1 restore node2 RAC Oracle Netbackup

tien86
Level 4
Partner

Hi,

I run Oracle RAC environemnt 11g2 NBU 7.6

Im having 2 problems takes me alot of time.

1,

I backup on node2 

RUN {
ALLOCATE CHANNEL ch00 
    TYPE 'SBT_TAPE';
ALLOCATE CHANNEL ch01 
    TYPE 'SBT_TAPE';
SEND 'NB_ORA_CLIENT=pe-pub2,NB_ORA_SID=pub,NB_ORA_POLICY=oradb,NB_ORA_SERV=poc1.lab.svtech';
BACKUP
    AS COMPRESSED BACKUPSET  
    $BACKUP_TYPE
    FORMAT 'bk_pub%u_s%s_p%p_t%t'
    DATABASE;

RELEASE CHANNEL ch00;
RELEASE CHANNEL ch01;

 

Then only Node2 can see the backupset in RMAN, while node1 see the backupset as EXPIRE state. 

 

Node1

backup piece handle=bk_pubcdpls76c_s397_p1_t861805772 RECID=324 STAMP=861805772

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=bk_pubcepls76c_s398_p1_t861805772 RECID=325 STAMP=861805773

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=bk_pubcfpls77g_s399_p1_t861805808 RECID=326 STAMP=861805811

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=bk_pubcgpls78c_s400_p1_t861805836 RECID=327 STAMP=861805836

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=arch_uchpls798_s401_p1_t861805864 RECID=328 STAMP=861805864

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=arch_ucipls798_s402_p1_t861805864 RECID=329 STAMP=861805864

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=arch_ucjpls79o_s403_p1_t861805880 RECID=330 STAMP=861805880

crosschecked backup piece: found to be 'EXPIRED'

backup piece handle=ctrl_uckpls7a4_s404_p1_t861805892 RECID=331 STAMP=861805895

Crosschecked 330 objects

 

Node2

backup piece handle=bk_pubcdpls76c_s397_p1_t861805772 RECID=324 STAMP=861805772

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=bk_pubcepls76c_s398_p1_t861805772 RECID=325 STAMP=861805773

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=bk_pubcfpls77g_s399_p1_t861805808 RECID=326 STAMP=861805811

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=bk_pubcgpls78c_s400_p1_t861805836 RECID=327 STAMP=861805836

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=arch_uchpls798_s401_p1_t861805864 RECID=328 STAMP=861805864

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=arch_ucipls798_s402_p1_t861805864 RECID=329 STAMP=861805864

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=arch_ucjpls79o_s403_p1_t861805880 RECID=330 STAMP=861805880

crosschecked backup piece: found to be 'AVAILABLE'

backup piece handle=ctrl_uckpls7a4_s404_p1_t861805892 RECID=331 STAMP=861805895

Crosschecked 330 objects

 

- Any configuration to help Node1 can see above backupset?

 

 

2 - In RAC Backup scenerio, with load-balance and failover VIP exist, I configure a script on 2 nodes 

3> ALLOCATE CHANNEL ch00   TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=pe-pub1-vip)' CONNECT='sys/oracle@pub1';
4> ALLOCATE CHANNEL ch01   TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=pe-pub2-vip)' CONNECT='sys/oracle@pub2';

In node1, I set ORACLE_SID=pub1 and Node2 ORACLE_SID=pub2

Then I can run backup script from node1 or node2. Then I can restore OK.

 

However, in policy, client I set is pe-pub1-vip and pe-pub2-vip. When I run policy from NBU Master It executes 2 jobs. If I remove pe-pub2-vip from client, the jobs from pe-pub2-vip failed with "NULL" status.

As I understand in best practices for Oracle RAC, Option 1 should be set only pe-pub1-vip in client. Option 2 is edit the script according to each database instances, so that 2 scripts would launch on 2 nodes but 1 will stop and 1 will run. 

I want to ask what steps to configure avoid this policy problem?

Thank you.

 

1 ACCEPTED SOLUTION

Accepted Solutions

tien86
Level 4
Partner

Hello,

I dicuss with my team and verify RMAN Catalog is not relate to the problem. Because in RAC environment, both node using same control file.

I found solution here which not mention in Admin Guide. 

http://www.symantec.com/business/support/index?page=content&id=TECH43187

View solution in original post

5 REPLIES 5

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

Where is your Oracle RMAN Catalog?

Hopefully you have a central RMAN Catalog and is not using NOCATALOG in the backup scripts?

Ensure both nodes point to the same RMAN Catalog.

For restores, you need to handle this as 'redirected restore' where you will authorize restores for all node members in NBU with altnames (or No.Restrictions). See NetBackup for Oracle Administrator's Guide  .

Option 2 is best to ensure successful backup. Also described in the manual.

tien86
Level 4
Partner

Hi,

With option using catalougue, I can restore on other node

If I backup from node2 and restore on node1 

 

I need to set NB_ORA_CLIENT=pe-pub2

SEND 'NB_ORA_CLIENT=pe-pub2,NB_ORA_SID=pub,NB_ORA_POLICY=oradb,NB_ORA_SERV=poc1.lab.svtech';

 

And the backupset are all available.

However my team still check if there a option for not using catalogue. And script also for load-balanced. 

tien86
Level 4
Partner

Hello,

I dicuss with my team and verify RMAN Catalog is not relate to the problem. Because in RAC environment, both node using same control file.

I found solution here which not mention in Admin Guide. 

http://www.symantec.com/business/support/index?page=content&id=TECH43187

Marianne
Moderator
Moderator
Partner    VIP    Accredited Certified

This is exactly what I said in my post above:

For restores, you need to handle this as 'redirected restore' where you will authorize restores for all node members in NBU with altnames (or No.Restrictions). See NetBackup for Oracle Administrator's Guide  .

tien86
Level 4
Partner

I already test option No.Restrictions file already and no work.

No.Restrictions is using with option restore with the same nodes while my question is restore using one node ( very different ) . It suit to the Solution 2 of the link and already works.