Please Read: Significant Update Planned, Migrating Forum Software This Month

See more
See less

File Share Witness Question

  • Filter
  • Time
  • Show
Clear All
new posts

  • File Share Witness Question


    I have a question about the FSW i'm hoping somebody can answer...

    Say i had a CCR environment that was spread over two sites as so:

    CCRNODE1 - Manchester
    CCRNODE2 - Sheffield

    I decide to put the FSW on a hub transport server in manchester for the CCR. If the Manchester sites fails, would CCRNODE2 automatically take control of the clustered mailbox? Bearing in mind it can't contact any server at Manchester.

    What i'm asking is does the FSW server need to be up for failover to occur?

    I've not got much experience in Microsoft clustering!


  • #2
    Re: File Share Witness Question

    Can anybody answer this for me?



    • #3
      Re: File Share Witness Question

      As long as 2 of the voting parties are up (Node 1, Node2, or the FSW), the cluster is ok. Kinda.

      If the FSW goes down, it does NOT mean a failover will occur, but you should fix the FSW ASAP to ensure that failover will occur if the active node goes offline.

      How to setup a cluster in server 2008:
      How to configure the FSW:
      A good (updated) note about NOT using CNAMEs for the FSW:


      • #4
        Re: File Share Witness Question

        Your scenario is kind of dangerouse
        Think of this scenario

        CCRNODE1 - Manchester
        CCRNODE2 - Sheffield
        FSW - Manchester

        When there is a problem with your ISP, and internet connection between Manchester and Sheffield goes down. CCRNODE2 will start up and become active when CCRNODE1 is also remain active. And each nodes will generate new data (logs and etc).
        You can't consolidate them.

        To do it right, you must have completely redundant internet services, and have FSW located in 3rd location, let's say Birmingham for example.

        CCRNODE1 - Manchester
        CCRNODE2 - Sheffield
        FSW - Birmingham

        Each server needs to be connected to 2 diffrent ISP and completely seperate switches, routers, gateways and etc.

        CCR ---ISP1--- FS ---ISP1--- CCR ---ISP1 --- CCR
        NODE1 ---ISP2--- W ---ISP2--- Node2 ---ISP2--- Node1

        -Incase ISP1 goes down on all sites, and ISP2 is still online, CCRNODE2 will not become active.

        -Incase ISP1 and ISP2 go down between CCRNODE1 and CCRNODE2, FSW will see both CCRNODE1 and CCRNODE2 so CCRNODE2 remain passive.

        -Incase ISP1 and ISP2 go down between CCRNODE1 and FSW, CCRNODE2 will see CCRNODE1, CCRNODE2 remain passive.

        -Incase ISP1 and ISP2 go down between CCRNODE2 and FSW, CCRNODE2 will see CCRNODE1, CCRNODE2 remain passive.

        Yes, it will be very unfortunate if all ISPs go down on all sites, that moment both CCRNODE1 and CCRNODE2 will be active and you will have a whole lot a different problems.

        My another recommendation would be setup CCR on Manchester on two different buildings....if you LAN is highly dependable, deploy 2 gateways on each buildings (possibly 2 diffrent ISPs), and set up SCR on Sheffield office.
        That will give you less headache, redundancy, and still be able to maintain backup email data on offsite (Sheffield), ready to go even if both buildings in Manchester goes up in flames. You need to be in Sheffield and make SCR to be active servers...... manually, if that happens however....
        Last edited by hyeongkim; 2nd April 2009, 18:04. Reason: clarify
        MCP, MCSA+messaging, MCDBA, OCA.