question

CatherineJaszewski-5685 avatar image
0 Votes"
CatherineJaszewski-5685 asked AndyDavid commented

Outlook Connectivity drops when one member of DAG goes down

I have an Exchange 2019 environment comprised of two mailbox servers (and an Edge Transport ‎server). The two mailbox servers are in one DAG which is working fine. However, when I take down ‎one of the servers in the DAG I lose Outlook client connectivity. I perform a Server Switchover before ‎taking the server down so the passive databases mount and activate properly. And the Outlook clients ‎are okay until I actually take the server down. ‎
Mail is still flowing (i.e. mail is not getting held up in queues either on the mailbox server or the Edge ‎Transport server) so I am confident the connectors are working properly. ‎
OWA is working fine both on the network and outside the network so I know the mailbox database is ‎working properly. ‎
I suspect the problem is with Autodiscover as the Config.XML file on the client does not change to ‎reflect the need to redirect to the other server. I haven’t changed any of my Virtual Directories since ‎creating the DAG (i.e. the virtual directory paths point to the servers not the DAG). And wondering if I ‎need to change the virtual directories to point to the CNO for the DAG instead of the Exchange server ‎itself. If this what I need to do resolve the Outlook connectivity issue, which virtual directories do I ‎need to change? ‎

office-exchange-server-administration
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

1 Answer

AndyDavid avatar image
0 Votes"
AndyDavid answered AndyDavid commented

What are you using for a client load balancer? :)

https://docs.microsoft.com/en-us/exchange/architecture/client-access/load-balancing?view=exchserver-2019

The virtual dir URLs on each server should all be the same and pointing to the Load Balancer pools ( or DNS round robin if using that)

P.S. That is not the same as the DAG CNO. Do not point clients to that. Its not supported!

· 10
5 |1600 characters needed characters left characters exceeded

Up to 10 attachments (including images) can be used with a maximum of 3.0 MiB each and 30.0 MiB total.

Hi Andy,

thank you for replying so quickly!
I I am not sure what internal load balancer we're using. I know we have F5 load balancer in the DMZ which directs traffic from the Internet. We have two IPs configured in an Priority group with the IP address from the Primary Exchange server set as high priority and IP address from the Secondary Exchange server set with a lower priority. The F5 load balancer performs health checks and if the primary IP is not available (i.e. server is down) traffic automatically gets sent to Secondary Exchange server.
But I think you are talking more about an internal load balancer and I do not recall setting up a load balancer pool for our Exchange environment. We do have one static IP in DNS that points to the Autodiscover service on the primary Exchange server.
Where would a Load Balancer pool be configured?

Please advise.
thank you!

0 Votes 0 ·
AndyDavid avatar image AndyDavid CatherineJaszewski-5685 ·

Yea, if you have an internal F5 avail, thats the way to go since the internal clients arent using the F5 in the DMZ.
F5 has lots of documentation on it, but really all you need is it to do is distribute the load nd mark a server down if 443 is not available for any reason. No need really for even session affinity.

You can also have the load balancer mark a server down during maintance :

To ensure that load balancers do not route traffic to a Mailbox server that Managed Availability has marked as offline, load balancer health probes must be configured to check <virtualdirectory>/healthcheck.htm, for example, <https://mail.contoso.com/owa/healthcheck.htm>;.

DNS Round Robin is really a poor way to load balance.




0 Votes 0 ·

Hi Andy,

I am a bit confused about how to describe my topology (i.e. Consolidated Datacenter, Single physical site with proxy....etc.)

I have two mailbox servers on two different subnets on the same network and located in one domain. There is one DAG comprised of the two mailbox servers. There is an Edge Transport server outside the network which is primarily used for Internet mail traffic (and ActiveSync). We are using an F5 load balancer hosted by a third party configured with the two IP addresses (one for each of the two mailbox servers) and these are configured in a Priority Group with the primary Mailbox server given high priority. If the primary IP address does not come back healthy (http request to port 443), traffic is directed to the IP address of the secondary mailbox server.

I am thinking this is considered a Consolidated Datacenter Model. Correct?

Please advise.

0 Votes 0 ·
AndyDavid avatar image AndyDavid CatherineJaszewski-5685 ·

Well, thats tough to say since its kind of a vague term :)

By the way, How does the Edge transport role apply to ActiveSync?

0 Votes 0 ·
Show more comments