HACMP auto-varyon - Aix

This is a discussion on HACMP auto-varyon - Aix ; hello! In clvefify.log we get : "One VG (not this one) set to autovaryonERROR: Volume Group oracle on Node node2 is configured to auto-varyon. FAIL" Configuration: AIX 5.3; HACMP 5.2, two node mutual take-over cluster hosting one Oracle resorce group, ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: HACMP auto-varyon

  1. HACMP auto-varyon

    hello!

    In clvefify.log we get :
    "One VG (not this one) set to autovaryonERROR: Volume Group oracle on
    Node node2 is configured to auto-varyon.
    FAIL"

    Configuration:
    AIX 5.3; HACMP 5.2, two node mutual take-over cluster hosting one
    Oracle resorce group, non-concurent shared VG based on EMC Symetrix
    storage. VG oracle is based on 6 hdiskpower# disks

    what was done:
    1. SAN reconfiguration - the path to hdisks were changed,
    2. rmdev by powermt check

    On node2 disapeard PVIDs on 2 disks. from now on we could not move
    resourse group to node2.

    we set up PVID by chdev -l hdiskpower# -a pv=yes on node2. The PVIDs
    are the same on both nodes.

    Then we tried to move resourses to node2 but no luck - the lazy update
    did not help
    We decided to Synchronize a Shared Volume Group Definition via C-SPOC.

    now we can move resource group oracle do node2

    now we have
    lsvg oracle

    QUORUM: 1
    AUTO ON: yes
    and the error above autovaryonERROR

    how can change chvg -a n ?? do we also must chage the QUORUM
    chvg -a n -Q n

    We would like to know detailed steps how to recover our cluster??
    we plan to

    stop HACMP services on both nodes
    on node2:
    importvg -y oracle
    varyonvg oracle
    chvg -a n -Q n oracle
    varyoffvg oracle
    start HACP services on both nodes
    test and check whether clverify succeeds or not

    do we have do the same on node1??

    I appreciate Your help
    Kind regards
    bizu


  2. Re: HACMP auto-varyon

    bizu wrote:
    > hello!
    >
    > In clvefify.log we get :
    > "One VG (not this one) set to autovaryonERROR: Volume Group oracle on
    > Node node2 is configured to auto-varyon.
    > FAIL"
    >
    > Configuration:
    > AIX 5.3; HACMP 5.2, two node mutual take-over cluster hosting one
    > Oracle resorce group, non-concurent shared VG based on EMC Symetrix
    > storage. VG oracle is based on 6 hdiskpower# disks
    >
    > what was done:
    > 1. SAN reconfiguration - the path to hdisks were changed,
    > 2. rmdev by powermt check
    >
    > On node2 disapeard PVIDs on 2 disks. from now on we could not move
    > resourse group to node2.
    >
    > we set up PVID by chdev -l hdiskpower# -a pv=yes on node2. The PVIDs
    > are the same on both nodes.
    >
    > Then we tried to move resourses to node2 but no luck - the lazy update
    > did not help
    > We decided to Synchronize a Shared Volume Group Definition via C-SPOC.
    >
    > now we can move resource group oracle do node2
    >
    > now we have
    > lsvg oracle
    >
    > QUORUM: 1
    > AUTO ON: yes
    > and the error above autovaryonERROR
    >
    > how can change chvg -a n ?? do we also must chage the QUORUM
    > chvg -a n -Q n
    >
    > We would like to know detailed steps how to recover our cluster??
    > we plan to
    >
    > stop HACMP services on both nodes
    > on node2:
    > importvg -y oracle
    > varyonvg oracle
    > chvg -a n -Q n oracle
    > varyoffvg oracle
    > start HACP services on both nodes
    > test and check whether clverify succeeds or not
    >
    > do we have do the same on node1??
    >
    > I appreciate Your help
    > Kind regards
    > bizu
    >

    Hi,

    I assume that the VG is in varyon state on node1, and autovaryon and
    Quorum are not set on node1. The VG is known on node2 but with wrong
    autovaryon and Quorum settings.

    If this is the case, you could try the following:
    just run chvg against the VG on node2 to set the correct parameters. You
    will get an error message, that the settings will not persist as the VG
    is not in varyon.
    You can ignore this message because the ODM is modified on node2 and the
    VGDA on disk already contains the correct settings due to node1.

    Then retry the verify.


    Regards
    Helmut

  3. Re: HACMP auto-varyon

    Thanks for Your comment.
    The oracle VG is working right now on node2. If I am right i can just
    change VG settings this way:
    Stop HACMP services on both nodes with forced option - all resources
    are available on node2 then
    chvg -a n oracle
    start HACMP
    or
    Stop HACMP services on both nodes with graceful option.
    on node2:
    varyonvg oracle
    chvg -a n oracle
    varyofvg oracle
    Start HACMP services

    please correct me if I am wrong.
    kind regards
    Bizu


  4. Re: HACMP auto-varyon

    bizu wrote:
    > Thanks for Your comment.
    > The oracle VG is working right now on node2. If I am right i can just
    > change VG settings this way:
    > Stop HACMP services on both nodes with forced option - all resources
    > are available on node2 then
    > chvg -a n oracle
    > start HACMP
    > or
    > Stop HACMP services on both nodes with graceful option.
    > on node2:
    > varyonvg oracle
    > chvg -a n oracle
    > varyofvg oracle
    > Start HACMP services
    >
    > please correct me if I am wrong.
    > kind regards
    > Bizu
    >

    That should be ok and is a safe way without tricks.

+ Reply to Thread