none
MapiExceptionRpcServerTooBusy

    Question

  • Hi All,

    I just went into production with an medium-sized Exchange 2010 environment (~1200 users) and I'm noticing a CAS issue and store.exe crashing under moderate load. The end user impact is Outlook disconnecting (2010, in cached and online mode) and OWA becoming unavailable (same message as error below). The error I'm receiving on th CAS server is below:

    Microsoft.Mapi.MapiExceptionRpcServerTooBusy: MapiExceptionRpcServerTooBusy: Unable to make connection to the server

    The environment:

    2 Physical boxes (12 cores each, 48GB RAM) running all roles (CAS,HT,MBX)

    Cas Array - physical HNLB running SNAT to Exchange Servers

    2 server DAG

    I expect it to be a max TCP or CAS connection limit (as the HNLB is SNAT-only) but haven't seen much in terms of settings. Bing has yet to give me much on the error... anyone else seen this under high load?

    Thanks,

    Luke

    Tuesday, August 03, 2010 11:39 PM

Answers

All replies

  • Have you run Best PRactice Analyzer from the toolbox?  Do you have BES Internal?  how about DC/GC servers in the site with Exchange?  What are the mailbox sizes?  What is the ratio of cached to online Outlook profiles?

    Sorry to be vauge but trying to isolate the possobilities.  Seems the hardware should be sufficient. 


    Sean McNeill Microsoft Gold Partner http://staterainfrastructure.blogspot.com/
    Wednesday, August 04, 2010 12:09 AM
  • No problem on the vagueness, here are my responses:

     

    Have you run Best PRactice Analyzer from the toolbox?  Yes, no issues.

    Do you have BES Internal?  No.

    how about DC/GC servers in the site with Exchange? 2 of then, 2008 R2. No performance issues/latency for GC lookups.

     What are the mailbox sizes?  ~400GB database for 1200 mailboxes.

    What is the ratio of cached to online Outlook profiles? High (95% cached, 5% online)

    Thanks,

    Luke

    Wednesday, August 04, 2010 5:30 AM
  • Did you tried MAGIC button - reboot exchange server, Hardware load balancer.

    Is this clear the problem immediately after you reboot, and it appear again after a few hours or days.

    How is CAS,MB,HT server performance? High?

     

    Just my 2cent, since current user size is 1200 users and keep on growing, it is recommended to split exchange multi roles so that the two server do not suffer bottleneck traffic and achieve better performance serve atomic purpose.

    Could this be bottleneck traffic where the server are failed to response this massive amount of concurrent traffic?


    Just a reference.

    http://www.networkworld.com/community/node/48015

    In these environments, you’d split the CAS/HT roles from the MBX roles and no longer have an “all in one” server.


    MT | MCITP ----- Please vote helpful or mark as answer if it's answered your question, this help us follow up the question status.
    Wednesday, August 04, 2010 7:59 AM
  • Luke,

    So sadly nothing you elaborated leads to a probable cause.

    What does the storage sub-system look like on the boxes?

    How frequent are the warnings/delays on the workstations?


    Sean McNeill Microsoft Gold Partner http://staterainfrastructure.blogspot.com/
    Wednesday, August 04, 2010 1:36 PM
  • Hi LK23,

    Any update for your issue?
    It sounds odd, I would suggest that you open a n ticket from MS.

    Regards!
    Gavin
    Monday, August 09, 2010 7:35 AM
  • This may be a bug with UR4. I have a premier case open and will post resolution when I have one. Thanks,

    Luke

    Monday, August 09, 2010 6:38 PM
  • Luke,

    Did you ever get a resolution to this problem?  We may be seeing the same issue on our 2010 environment.

    Thanks!

    //Tony

    Tuesday, January 04, 2011 5:18 PM
  • Sorry for the delay - yes, it was a bug with the update rollup and fixed in SP1.
    Monday, January 10, 2011 6:35 PM