none
After server failure, clients won't reconnect

    Question

  • Hi guys,

    Having issues with our Outlook 2010 clients in our pre-pilot deployment.

    If mailboxes of the test users resides in DB01 and we restart the server currently holding the active database, the clients gets disconnected but it takes up to 1 hour before they are connected again, or if we restart Outlook.

    If we do a database "switchover" from the shell or from within EAC the clients gets disconnected for 2-3 secs, then they connect perfectly.

    We have the following Exchange Server 2013 environment:

    - 3 Multirole servers hosting active/passive databases

    - 1 multirole server hosting passive lagged copies

    Hardware load balancer / HTTPS traffic for autodiscover.company.com and mail.company.com

    - LB method: Round Robin (between server 1-3, not the lagged one)

    - Persistance: none

    - Monitor: SSL TCP, 5 sec interval, 2 sec time-out, 3 retries, 30 sec downtime

    While we initiate a server reboot and the databases are mounted on another server (we monitor this via powershell), I can at the same time in outlook under Connection Status see that it lost some proxy servers and dont reconnect automatically.

    If I press Reconnect, all proxyservers are removed and that list is empty...

    So.. Howcome I can initiate a databaswe switchover and things are running perfect but when initiating a serverfailure it doesnt work. Is this HLB related, outlook related or Exchange related?

    Help appreciated!



    • Edited by [fredrik] Tuesday, June 25, 2013 11:48 AM
    Tuesday, June 25, 2013 7:47 AM

All replies

  • Hi

    I guess it is related to client

    Did that happen to all clients or just a few of them.

    If just for some users, please have a check this thread

    http://social.technet.microsoft.com/Forums/en-US/d9aeeaeb-95e1-468b-8dc7-e23066b63afa/clients-will-not-reconnect-after-failover

    Perhaps some attribute on users is pointing the wrong direction. try to move mailboxes for teh 10 users to another DB and see what happens, by moving mailboxes, attributes will be updated on user object in AD, also configure a clientaccessarray object with an A record in DNS to direct your clients

    And if The client could work well if the profile recreated, that means the outlook end could not update the connnection information using aurodiscover service. 

    Cheers

    If you have any feedback on our support, please click here


    Zi Feng
    TechNet Community Support

    Wednesday, June 26, 2013 6:31 AM
  • Hi,

    Since this is Exchange Server 2013, we don't have a CAS array any more. It happens on all our test clients (still in pre-pilot phase)

    One thing I noticed though is that RPCClientAccessServer (that was used in legacy exchange versions on the databases) is configured with another Mailbox server that the current one holding the active copy... how come? is this even used now days?


    • Edited by [fredrik] Wednesday, June 26, 2013 7:08 AM spelling
    Wednesday, June 26, 2013 7:07 AM
  • any more hints?

    Tuesday, July 02, 2013 6:33 AM
  • Sorry for bumping this thread, but still in need of some help!
    Tuesday, July 30, 2013 8:18 AM