PowerHA for AIX

 View Only
Expand all | Collapse all

POWER HA 7.1 and SVC Synchrnous Mirror

  • 1.  POWER HA 7.1 and SVC Synchrnous Mirror

    Posted Thu December 25, 2014 01:46 PM

    Originally posted by: polani


    Guys

     

    We have a POWER HA 7.1 standard edition implemented with Enhanced concurrent mode VG(s) and shared disks are replicated to DR site using SVC synchronous mirror.

     

    The Lpar at DR site is a standalone LPAR. We did a successful failover to DR recently , by breaking SVC synchronous mirrors and getting all filesystems mounted at DR site. However the volume group at DR site was also a enhanced concurrent vg ( as expected ) and varied on manually.

     

    We did not make any data changes on DR site during DR excersice , however the question is what if we change the data contents of filesystems and get them unmounted , volume group varied off at DR site and then reverse replication direction from DR to production site ? Would it impact any thing on production POWER HA clusters? Off course we would stop POWER HA clusters before reversing data replication direction however my real question is as we varied on this ECVG in manual mode at DR site , once we will replicate it back to production site then VG's changed characteristics would also probably replicated to production site and we might face issues while starting POWER HA clusters ... anybody did this kind of replication direction reversal while using POWER HA , please guide ... many thanks

     

     



  • 2.  Re: POWER HA 7.1 and SVC Synchrnous Mirror

    Posted Mon December 29, 2014 10:48 AM

    Originally posted by: POWERHAguy


    PowerHA Standard Edition is replication and really data unaware. So what you describe does not have to cause any grief to the local cluster. However, IF you make LVM/JFS related changes it will cause some minor grief. But that's not too different than if you did the same locally either. A verify/sync before starting cluster again locally can usually take care of that anyway.

    I have a local customer that we are doing pretty much what sounds like the exact same thing with v7000. Two node local cluster for HA, and a third system/lpar at a DR site replicated to for manual recovery.

     

    Shawn Bodily
    Senior Consultant - Clear Technologies

    PowerHAguy on Youtube and Twitter