This discussion is archived
1 2 Previous Next 16 Replies Latest reply: Mar 4, 2013 5:25 PM by yxes2013 RSS

Standby Failover

yxes2013 Newbie
Currently Being Moderated
Hi all,

I am testing a failover to my standby database. I want to make my standby as the primary. But when I try to startup open the standby, I got message error saying the database was not a standby :(

What do you think did I miss?

Thanks,

zxy
  • 1. Re: Standby Failover
    Fran Guru
    Currently Being Moderated
    when you make the backup of primary you must have a controlfile for standby "rman>BACKUP CURRENT CONTROLFILE FOR STANDBY"

    check this for other solution:
    http://odenysenko.wordpress.com/2012/07/02/quick-solution-for-ora-01665/
  • 2. Re: Standby Failover
    yxes2013 Newbie
    Currently Being Moderated
    Thanks frannie :)
  • 3. Re: Standby Failover
    Hemant K Chitale Oracle ACE
    Currently Being Moderated
    It depends on how the "standby" was built and how it is being updated.

    The old old method (aka "manual standby") and the method for Oracle Standard Edition don't use DataGuard. They don't use a Standby Controlfile either.


    Hemant K Chitale
  • 4. Re: Standby Failover
    yxes2013 Newbie
    Currently Being Moderated
    Hi all,

    If I am sending archive logfiles to our standby db(not-dataguard) every 5 minutes and apply it continuously.
    Does this mean that if disaster strikes, i will lose 5 minutes worth of transaction?

    How do you manage your standby such that minimal data loss is attained?

    Please share your current setup :(


    Thanks a lot,

    zxy
  • 5. Re: Standby Failover
    Hemant K Chitale Oracle ACE
    Currently Being Moderated
    Yes, you would lose 5minutes of transactions.


    IF you want maximum protection you would configure MAXIMUM PROTECTION mode and SYNC so that transactions are immediately written to the Standby. This is feasible if the network latency to the Standby is not very high -- otherwise the latency will "kill" the performance of commits.
    Also, in MAXIMUM PROTECTION the risk to the Primary is that transactions on the Primary will halt if the Standby is down or the network link to the Standby is down.


    Hemant K Chitale
  • 6. Re: Standby Failover
    Shivananda Rao Guru
    Currently Being Moderated
    Hello,

    Refer this on how to perform a failover http://shivanandarao.wordpress.com/2012/08/28/dataguard-failover/

    Which edition and version of database are you using ? If its an enterprise edition, then why do you choose to ship the archives manually every 5 minutes ?
    Oracle automatically ships it and applies it to the standby.


    Regards,
    Shivananda
  • 7. Re: Standby Failover
    CKPT Guru
    Currently Being Moderated
    I got message error saying the database was not a standby :(
    SQL> Select controlfile_type, database_role from v$database;


    >
    If I am sending archive logfiles to our standby db(not-dataguard) every 5 minutes and apply it continuously.
    Does this mean that if disaster strikes, i will lose 5 minutes worth of transaction?
    >

    Yes, You will have the lost of data if it is not Data Guard and also not an real-time apply.

    >
    How do you manage your standby such that minimal data loss is attained?
    >

    As my personal experience, To keep archive location for both primary and standby in shared location
  • 8. Re: Standby Failover
    yxes2013 Newbie
    Currently Being Moderated
    As my personal experience, To keep archive location for both primary and standby in shared location
    Wow! brilliant idea dear :)

    This can be done even if the disk is internal to the primary right? I just remote mount it as read only at the standby?

    My question is , can the two database create a "deadlock" on a certain archivelog? In which the primary db is still offloading the archivelog and the standby db simultaneously apply it?
    Will it not corrupt the standby incase the connection is lost?
  • 9. Re: Standby Failover
    CKPT Guru
    Currently Being Moderated
    yxes2013 wrote:
    As my personal experience, To keep archive location for both primary and standby in shared location
    Wow! brilliant idea dear :)

    This can be done even if the disk is internal to the primary right? I just remote mount it as read only at the standby?
    Internal of primary means, only attached to Primary? In that point No.
    I mean to say shared storage it can be SAN/NAS, which is attached to both primary and standby so that both can access.
    My question is , can the two database create a "deadlock" on a certain archivelog? In which the primary db is still offloading the archivelog and the standby db simultaneously apply it?
    Will it not corrupt the standby incase the connection is lost?
    Nope, Archive log file is online dumped log file :) so primary wont use once if it is archived.
    But i can say, you have to ensure the RMAN deletion job should not run on primary after they applied on standby database.
    So the recovery procedure should run continuously and RMAN archive job(backup + deletion) with a 4-6 hours gap. Then there will be no conflict. Even on standby once it is used and applied , it wont use again. So no worries.
  • 10. Re: Standby Failover
    yxes2013 Newbie
    Currently Being Moderated
    Internal of primary means, only attached to Primary? In that point No.
    If I am cost cutting, why is a remote mount disk not possible to do? :(
    What is the issue?

    When in fact the standby only reads the disk?
    Nope, Archive log file is online dumped log file so primary wont use once if it is archived.
    Yes I know dear, :) but what mean is which the primary is currently offloading the online redo log to the archive log format (on-going and not yet finished), The standby already caught it and apply it on its database. Something like they are toggling on the file.
  • 11. Re: Standby Failover
    CKPT Guru
    Currently Being Moderated
    yxes2013 wrote:
    Internal of primary means, only attached to Primary? In that point No.
    If I am cost cutting, why is a remote mount disk not possible to do? :(
    Then ensure you are able to transport archives as soon as they generated on primary. No other options.
    Or you can see the alternate option with good pack http://www.dbvisit.com/products/dbvisit_standby_database_for_oracle_disaster_recovery
    When in fact the standby only reads the disk?
    Nope, Archive log file is online dumped log file so primary wont use once if it is archived.
    Yes I know dear, :) but what mean is which the primary is currently offloading the online redo log to the archive log format (on-going and not yet finished), The standby already caught it and apply it on its database. Something like they are toggling.
    Am not saying to put Online redo log files in shared location, Only archives location what ever you mentioned in "LOG_ARCHIVE_DEST_n"
    Moreover there is nothing to do for standby with primary online redo log files, when there is no connection between them.
    Got it what am saying?
  • 12. Re: Standby Failover
    yxes2013 Newbie
    Currently Being Moderated
    Hi Chkpt,

    I just realize the shared disk defeats the purpose of standby backup :(
    As standby should be located at remote side far from the primary such that incase there is tsunami or strong earthquake that collapses the building, the standby would not be affected :D
  • 13. Re: Standby Failover
    CKPT Guru
    Currently Being Moderated
    yxes2013 wrote:
    Hi Chkpt,

    I just realize the shared disk defeats the purpose of standby backup :(
    As standby should be located at remote side far from the primary such that incase there is tsunami or strong earthquake that collapses the building, the standby would not be affected :D
    Yes, If data center is too far, Then you are correct.
    Now only options are
    1) Fast SCP of archived files from primary to standby using jobs
    2) Use Data Guard
    3) use Dbvisit

    I do not have options more than these ;)
  • 14. Re: Standby Failover
    Hemant K Chitale Oracle ACE
    Currently Being Moderated
    To keep archive location for both primary and standby in shared location
    It becomes a SPOF (Single Point of Failure)

    It won't work across long distances -- which is what DR / Standby Sites are generally expected to be.

    You would use this implementation for a Reporting environment that uses DataGuard architecture.


    Hemant K Chitale
1 2 Previous Next

Legend

  • Correct Answers - 10 points
  • Helpful Answers - 5 points