+ Reply to Thread
Results 1 to 7 of 7
  1. Senior Member
    Join Date
    Nov 2007
    Location
    Norway
    Posts
    232

    Certifications
    MCTS, MCSA
    #1

    Default Upgrading DCs from 2003 to 2008 R2 with Exchange 2010 in the domain

    What should be the course of action? Does one need to take something special into consideration, regarding the Exchange part? Or is it just simply joining the 2008 R2 DCs to the domain, and then demoting the 2003 ones, like one does normally when doing this type of upgrade?

    Does the Exchange server have SRV records in DNS?
    Reply With Quote Quote  

  2. SS -->
  3. Nidhoggr, the Net Serpent Claymoore's Avatar
    Join Date
    Nov 2007
    Location
    FL
    Posts
    1,622

    Certifications
    AWS Architect, MCSEx3, MCITPx6, MCTSx17
    #2
    Exchange has AD properties but there's nothing special you need to do when installing new domain controllers. Exchange has an AD topology service that regularly checks for changes to AD service availability so it will see the new servers come online and the old ones drop off, and change the Global Catalog queries as necessary. Just be sure that you have a GC or two available in each AD site that has an Exchange server.

    It's possible that you have a SRV record for Autodiscover, but not probable. SRV records are one possible way of locating Autodiscover information, but that method is being deprecated.
    Reply With Quote Quote  

  4. Senior Member
    Join Date
    Nov 2007
    Location
    Norway
    Posts
    232

    Certifications
    MCTS, MCSA
    #3
    Isn't there supposed to be a MX record for the Exchange server? I can not find such a record in my DNS zones.
    Reply With Quote Quote  

  5. Senior Member
    Join Date
    Nov 2007
    Location
    Norway
    Posts
    232

    Certifications
    MCTS, MCSA
    #4
    I have actually performed this upgrade, and now I'm struggling with the Exchange Server loosing connection with the DCs and GCs sporadically. Restarting the Exchange Server resolves the problem. But after a while the problem recurs.

    I get the following error messages in Event Viewer on the application log of the Exchange Server

    Process STORE.EXE (PID=6772). All Global Catalog Servers in forest DC=xx,DC=xx are not responding:
    DC001.xx.xx
    DC002.xx.xx

    Process MSEXCHANGEADTOPOLOGYSERVICE.EXE (PID=1348 ). All Global Catalog Servers in forest DC=xx,DC=xx are not responding:
    DC001.xx.xx
    DC002.xx.xx

    Process MAD.EXE (PID=5500). All Global Catalog Servers in forest DC=xx,DC=xx are not responding:
    DC001.xx.xx
    DC002.xx.xx



    I get the following error when the processes loose connection with the DCs/GCs, The Microsoft Exchange Information Store service terminated with service-specific error %%-2147467259.

    These messages started to appear after I demoted the old 2003 DCs. The course of action was simple, run adprep, promote 2008 R2 servers to DC/DNS servers, move all FSMO roles to DC001, demote the old 2003 DCs as DC/DNS servers. Remove the 2003 DCs from the domain.

    Everything else is working fine. Exchange is the only service that is causing problems. There is only one site, but two different subnets, at two different locations. No subnets have been defined in AD sites and services, as there is only one site (Default-First-Site). I have checked with Nltest and the Exchange server, the DCs and clients all belong to the same site.

    I have run Get-Exchangeserver -status | fl on the Exchange server, and there are no statically defined GCs. And the GCs and DCs reported as current are correct (the new Server 2008 R2 ones).

    Event 2080 gives the following output when everything is working fine

    Process MAD.EXE (PID=5556). Exchange Active Directory Provider has discovered the following servers with the following characteristics:
    (Server name | Roles | Enabled | Reachability | Synchronized | GC capable | PDC | SACL right | Critical Data | Netlogon | OS Version)
    In-site:
    DC001.xx.xx CDG 1 7 7 1 0 1 1 7 1
    DC002.xx.xx CDG 1 7 7 1 0 1 1 7 1
    Out-of-site:

    Process MSEXCHANGEADTOPOLOGYSERVICE.EXE and Process STORE.EXE gives the same output.

    Do you know what could be causing this?
    Reply With Quote Quote  

  6. Nidhoggr, the Net Serpent Claymoore's Avatar
    Join Date
    Nov 2007
    Location
    FL
    Posts
    1,622

    Certifications
    AWS Architect, MCSEx3, MCITPx6, MCTSx17
    #5
    You need to define your subnets in AD sites and services. Exchange is site-aware and will be looking for a GC in the same site, as well as making sure there are CAS and Hub servers in the same site as the Mailbox server.
    Reply With Quote Quote  

  7. Senior Member
    Join Date
    Nov 2007
    Location
    Norway
    Posts
    232

    Certifications
    MCTS, MCSA
    #6
    Thanks, I have defined the subnets, and will see how it goes. But I have another strange problem. I have an AD-Integrated primary DNS zone, which is called webmail. FQDN, and it contains just a cname record pointing to the exchange server. It works fine resolving webmail. FQDN to the ip of the exchange server, and the Name servers on the name server tab are the two new DCs.

    But the SOA record has one of the old DCs defined as the primary server. If I try to change the SOA record, I get the following error Message

    "The SOA record cannot be updated. Node is a CNAME DNS record."

    TTL for the SOA record is 1 hour. I've tried to do the following

    1. Add a random host record to the zone, and then remove the record. This will increment the Serial Number, but the primary server in SOA stays the same.

    2. Tried to just increment the serial number, but then I receive the same error message as when trying to manually change the SOA record.

    3. Tried to make the zone a regular (non AD integraged) zone, but still I am not able to change the SOA record.

    Any suggestions on resolving this problem, or is the only solution to delete the zone and recreate it?
    Reply With Quote Quote  

  8. Nidhoggr, the Net Serpent Claymoore's Avatar
    Join Date
    Nov 2007
    Location
    FL
    Posts
    1,622

    Certifications
    AWS Architect, MCSEx3, MCITPx6, MCTSx17
    #7
    Why do you have a subdomain for one record? The webmail host record (A or CNAME) belongs as a host record in the parent zone. Only reason I can think of having a separate zone is to do some funky DNS replication, which you weren't doing if the zone were AD integrated. Delete the zone and add the host record to the parent zone.
    Reply With Quote Quote  

+ Reply to Thread

Social Networking & Bookmarks