No announcement yet.

Question about Hub Transport Redundancy / Resiliency

  • Filter
  • Time
  • Show
Clear All
new posts

  • Question about Hub Transport Redundancy / Resiliency

    Hi all,

    I am trying to do some research on a question to better understand Hub Transport.

    Right now we are a multisite organization with a Hub Transport server at each location. Each of our sites have their own internet connection, and are all connected via some type of P2P connection independent of the internet access. The hub transports successfully send and receive mail for the organization.

    We have recently set up all this up to be more redundant, and yesterday came across our first real world test of the redundancy.

    One of our locations lost internet access thanks to AT&T having a router interface blow in one of their data centers. Our internet access at that location was down for 4 hours.

    In that time frame, the Hub Transport server at that location built up outgoing messages in its queue, with the expected failure message that it could not resolve or contact an SMTP server for the various domains.

    My understanding is that Active Directory and Exchange will automatically detect and failover transport services to another transport server if the actual Hub Transport server fails. However, I don't see anything written about how AD and Exchange handle outgoing mail on an HT when it all it does is lose its connection to the outside world.

    I tried removing the HT server with no internet access out of the send connector, and that seemed to keep NEW mail from being built up in the queue. However, the old mail was stuck in the queue of that transport server.

    Admittedly, I was rather busy putting out lots of fires thanks to AT&T, so I didn't get the opportunity to troubleshoot or play with things the way I wanted to. But how is it SUPPOSED to work? Is there something I can do to make that built up mail move to a queue on a different transport server?

    Any and all help is much appreciated. Thanks!

  • #2
    Re: Question about Hub Transport Redundancy / Resiliency

    I thought I would post a reply with the solution I found, and then an additional question...

    Clearing out the queue and routing that mail through a different HT server is as simple as removing the HT server with the internet problem from the default send connector, and then temporarily adding a second send connector for that HT server to route mail to another exchange server as a smart host. When you do that, and restart your transport service, the queue flushes out and all its mail moves to the other HT transport server you specified.

    So my additional question is... How does one make this an automatic failover? I understand that you can do weighted send connectors, but how does that work exactly? Does the HT try the first connector in its queue, and if it fails the first time, it tries the second? Or does the connector go through multiple retries based on the settings of the server before it tries another connector?


    • #3
      Re: Question about Hub Transport Redundancy / Resiliency

      Do you only have one Send Connector then?
      I am surprised that worked for you because the FQDN value wouldn't be correct for one or other of the sites.
      The way that I would do it is to setup a Send Connector for each location, with the server/s in that location listed as the source. Make them all the same cost and Exchange will use them all.

      Simon Butler
      Exchange MVP

      More Exchange Content:
      Exchange Resources List:
      In the UK? Hire me:

      Sembee is a registered trademark, used here with permission.


      • #4
        Re: Question about Hub Transport Redundancy / Resiliency

        That's a good point that I didn't think about, as our firewalls
        Rewrite the fqdn of outbound smtp traffic from our servers at each location.

        It works as is, but I'll change it for sure.