Failover without state. - Weblogic

This is a discussion on Failover without state. - Weblogic ; If we have a web service sitting on top of say stateless session beans, and we have a clustered Weblogic environment running on say solaris servers with an oracle database, if the app server fails during a particular web service ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: Failover without state.

  1. Failover without state.


    If we have a web service sitting on top of say stateless session beans, and we
    have a clustered Weblogic environment running on say solaris servers with an oracle
    database, if the app server fails during a particular web service thread, what
    would be transferred to a secondary server?

    Since there is no session state (by nature of a web service - you probably wouldn't
    have session state) is just the the state of the thread (including db connections?)
    up to the failure replicated over to the secondary server allowing the web service
    to continue to completion, or does the web service return unsuccessfully?

    What should I read to get up to speed on this? Thanks.

  2. Re: Failover without state.


    If you choose not to deploy session failover, or state replication, then you are
    simply providing a service/application that will stay up....but does not garuntee
    session replication.

    So in affect, if a server or instance fails, then the user may get an error, but
    would then be directed to another server/instance within the cluster. Obviously,
    without replication, he/she may lose a shopping cart or saved session info and
    will need to reauthenticate/start again.

    My advise would be to always use replication, its a good feature and is there
    to be used! If you have paid for it, then use it ?? As long as session sizes stay
    low, then there is just a small overhead if you have a larger heap.

    In term of documentation, search from the www.bea.com site for failover. and you
    be presented with loads of docs!!
    Good Luck
    "scohan" wrote:
    >
    >If we have a web service sitting on top of say stateless session beans,
    >and we
    >have a clustered Weblogic environment running on say solaris servers
    >with an oracle
    >database, if the app server fails during a particular web service thread,
    >what
    >would be transferred to a secondary server?
    >
    >Since there is no session state (by nature of a web service - you probably
    >wouldn't
    >have session state) is just the the state of the thread (including db
    >connections?)
    >up to the failure replicated over to the secondary server allowing the
    >web service
    >to continue to completion, or does the web service return unsuccessfully?
    >
    >
    >What should I read to get up to speed on this? Thanks.



  3. Re: Failover without state.

    "Greg OReilly" wrote:
    > In term of documentation, search from the www.bea.com site for failover


    For WebLogic Server 8.1, try this document first:

    http://e-docs.bea.com/wls/docs81/cluster/failover.html

    For WLS 7.0, there's a pretty good section on failover support within
    a cluster from the book, "BEA WebLogic 7 Server Administration" (ISBN
    0-07-222316-2). See chapter 8, page 232.


    Brian J. Mitchell
    Systems Administrator, TRX
    email: brian.mitchell@trx.com
    office: +1 404 327 7238
    mobile: +1 678 283 6530

    ----------------------------------------------------------------------
    PGP key: http://pgp.mit.edu:11371/pks/lookup?...rch=0x8BEE2240
    Fingerprint: 3A84 A5D6 EE26 BC10 E525 1905 6FF0 0B6F 8BEE 2240
    ----------------------------------------------------------------------

+ Reply to Thread