Forum Discussion

avertelis's avatar
avertelis
Level 2
7 years ago

Backups for Oracle RAC running twice

Hi All, We having issues with Oracle RAC invirovment backups, it's always dublicating. It was working, and did load balance between nodes, but i did created new policy with same configuration and no...
8
Oracle RAC
Windows Server 2008 R2
  • RiaanBadenhorst's avatar
    RiaanBadenhorst
    7 years ago

    Hi,

    The script you attached won't load balance. For that to happen you'd need to allocate channels using specific instances name and logins. Example below

    ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

    ALLOCATE CHANNEL ch01 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

    ALLOCATE CHANNEL ch02 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

    ALLOCATE CHANNEL ch03 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

    ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

    ALLOCATE CHANNEL ch01 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

    ALLOCATE CHANNEL ch02 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

    ALLOCATE CHANNEL ch03 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

    What you do then is make another policy that will kick off the backups by connecting to the SCAN. In that policy you don't need a application backup schedule as it will connect and initiate the backup script. Inside the backup script you'll list the policy name of the other policy (NB_ORA_POLICY=test11g1) as shown above. 

    Because the channels are allocated as shown above it will make the appropriate channels on each node and then send the data to the media server.

    Hope that helps.

RiaanBadenhorst's avatar
RiaanBadenhorst
Level 6
7 years ago

Hi,

The script you attached won't load balance. For that to happen you'd need to allocate channels using specific instances name and logins. Example below

ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

ALLOCATE CHANNEL ch01 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

ALLOCATE CHANNEL ch02 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

ALLOCATE CHANNEL ch03 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

ALLOCATE CHANNEL ch00 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

ALLOCATE CHANNEL ch01 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

ALLOCATE CHANNEL ch02 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb1-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g1';

ALLOCATE CHANNEL ch03 TYPE 'SBT_TAPE' PARMS='ENV=(NB_ORA_CLIENT=rwaixtdb2-vip,NB_ORA_POLICY=test11g1)' CONNECT='oraback/ora10gr2bk@test11g2';

What you do then is make another policy that will kick off the backups by connecting to the SCAN. In that policy you don't need a application backup schedule as it will connect and initiate the backup script. Inside the backup script you'll list the policy name of the other policy (NB_ORA_POLICY=test11g1) as shown above. 

Because the channels are allocated as shown above it will make the appropriate channels on each node and then send the data to the media server.

Hope that helps.

  • avertelis's avatar
    avertelis
    Level 2
    7 years ago

    Can you send me full script file?

    • RiaanBadenhorst's avatar
      RiaanBadenhorst
      Level 6
      7 years ago

      Hi,

      Sorry i only picked that piece from an email i sent a client a few years ago.

      You can check the NetBackup Oracle Guide (RAC section) for more information.

      • avertelis's avatar
        avertelis
        Level 2
        7 years ago

        Thanks Riaan,

        I created second policy wich is pointing scan name and kicking policy with two VIP names. When i starting backup job, first node connection failed, second node is backing up. Here is the log:

        Aug 9, 2017 9:24:48 AM - Info nbjm (pid=7868) starting backup job (jobid=464643) for client ch-cppdb-p01-vip, policy CH-CPPDB_RAC_RMAN_ARCH, schedule Default-Application-Backup

        Aug 9, 2017 9:24:48 AM - Info nbjm (pid=7868) requesting STANDARD_RESOURCE resources from RB for backup job (jobid=464643, request id:{7D24B411-0C63-42C8-A840-5D575930DEA4})

        Aug 9, 2017 9:24:48 AM - requesting resource NBU_Basic

        Aug 9, 2017 9:24:48 AM - requesting resource bku01.NBU_CLIENT.MAXJOBS.ch-cppdb-p01-vip

        Aug 9, 2017 9:24:48 AM - requesting resource bku01.NBU_POLICY.MAXJOBS.CH-CPPDB_RAC_RMAN_ARCH

        Aug 9, 2017 9:24:49 AM - granted resource  bku01.NBU_CLIENT.MAXJOBS.ch-cppdb-p01-vip

        Aug 9, 2017 9:24:49 AM - granted resource  bku01.NBU_POLICY.MAXJOBS.CH-CPPDB_RAC_RMAN_ARCH

        Aug 9, 2017 9:24:49 AM - granted resource  MediaID=@aaaat;Path=I:\Basic���Backups;MediaServer=bku01

        Aug 9, 2017 9:24:49 AM - granted resource  NBU_Basic

        Aug 9, 2017 9:24:49 AM - estimated 0 kbytes needed

        Aug 9, 2017 9:24:49 AM - Info nbjm (pid=7868) started backup (backupid=ch-cppdb-p01-vip_1502288689) job for client ch-cppdb-p01-vip, policy CH-CPPDB_RAC_RMAN_ARCH, schedule Default-Application-Backup on storage unit NBU_Basic

        Aug 9, 2017 9:24:49 AM - started process bpbrm (pid=2620)

        Aug 9, 2017 9:24:50 AM - Info bpbrm (pid=2620) ch-cppdb-p01-vip is the host to backup data from

        Aug 9, 2017 9:24:50 AM - Info bpbrm (pid=2620) reading file list for client

        Aug 9, 2017 9:24:51 AM - connecting

        Aug 9, 2017 9:24:53 AM - Info bpbrm (pid=2620) listening for client connection

        Aug 9, 2017 9:24:59 AM - Info bpbrm (pid=2620) INF - Client read timeout = 14400

        Aug 9, 2017 9:31:49 AM - Error bpbrm (pid=2620) listen for client protocol error - couldn't accept from data socket, An operation was attempted on something that is not a socket.  (10038)

        Aug 9, 2017 9:31:49 AM - Info dbclient (pid=0) done. status: 25: cannot connect on socket

        Aug 9, 2017 9:31:49 AM - end writing

        cannot connect on socket  (25)