SQL Server Performance

SQL Server Configuration Manager IP addresses list

Discussion in 'SQL Server 2005 Clustering' started by bozgeyik, Jul 2, 2007.

  1. bozgeyik New Member

    We have a 2-node SQL 2005 cluster which hosts two (named) instances at the same time.

    In SQL Server Configuration Manager, there is an IP addresses list for TCP/IP protocol for each DB instance under SQL Server 2005 Network Configuration tab.

    I noticed that the IP of that specific instance does not exist in the list. However, it is correctly set by cluster admin and can be seen via ipconfig, netstat etc.

    What would be the reason?

    The default setting for DB instances to listen on dynamic ports. I changed the configuration to listen to static ports.

    The strange thing is that, although the IP of that instance is not in the list and therefore cannot be changed, it still works correctly, i.e. it listens to the static port although it was listening on dynamic ports before I changed the setting for other IPs. How can this be?

    Another strange thing is that for one instance its IP is not listed at all and for the other instance, the IP is in the list of the other instance.

    Some stranger thing is that this behavior repeats itself for a totally different cluster which has the same characteristics, i.e. two-nodes, two instances.

    Everything works fine now, but I do not how this happens. Does anybody have an idea how such?

  2. satya Moderator

    Have you restarted SQL browser service after changing the ports from dynamic to static?

    Satya SKJ
    Microsoft SQL Server MVP
    Writer, Contributing Editor & Moderator
    http://www.SQL-Server-Performance.Com
    This posting is provided AS IS with no rights for the sake of knowledge sharing. Knowledge is of two kinds. We know a subject ourselves or we know where we can find information on it.
  3. bozgeyik New Member

    Sorry for the delayed reply. I was out of office for a week.

    Yes I did restart the browser service (and all the others as well).

    the IP list for the SQL cluster is as follows:

    x.x.x.a: cluster IP
    x.x.x.b: MSDTC IP
    x.x.x.c: instance1 IP
    x.x.x.d: instance2 IP
    x.x.x.e: server IP (node1)
    x.x.x.f: server IP (node2)
    and the IP list from the SQL server configuration manager run on node1 is as follows:

    when you select instance1 and view the IP addresses, you only have x.x.x.a and x.x.x.b and x.x.x.e included in the list among with the other irrelevant IPs (like other interfaces).

    when you select instance2 and view the IP addresses, you have x.x.x.a, x.x.x.b, x.x.x.c and x.x.x.e. That is, you have an instance IP now but it is the wrong instance.

    When you go to the other node and view the listings, you have exactly the same listing. That is, the server IP is again x.x.x.e (not x.x.x.f). This would have been contributed to the fact that the resources are on the node1. But, no... I moved all the resources to the other node successfully and restarted all the services including sql browser. When I view the listing again, it is still exactly the same.

    Where does that listing come from? why the records are not updated at all? Once again, everything works smoothly and I have all the IPs available in the list obtained by running ipconfig or netstat.

    thanks,

    Bora

  4. satya Moderator

    Is this an active/active or active/passive?
    I might need to look to reproduce at my end for further confirmation.

    Satya SKJ
    Microsoft SQL Server MVP
    Writer, Contributing Editor & Moderator
    http://www.SQL-Server-Performance.Com
    This posting is provided AS IS with no rights for the sake of knowledge sharing. Knowledge is of two kinds. We know a subject ourselves or we know where we can find information on it.
  5. bozgeyik New Member

    The sql cluster works in active/passive mode.

    Actually we have two clusters both working in active/passive mode. The aforementioned behaviour is the same in two clusters. The only difference between the clusters are:
    1) one is with SQL 2005 SP1 and the other is with SP2 (for some specific reason)
    2) server machine hardware are different. HP Proliant BL460c and HP Proliant DL585
    3) same disk controllers (MSA1000) and fiber switches on both clusters. MSA1000 works in active/active mode to test its performance as there are two instances which use different sets of disks, both have the same firmware. Only difference is the disk number and disk sizes.

    thanks for your efforts...

    Bora

Share This Page