DAG Exchange messages headers

Question

DAG Exchange messages headers

Step to IT 125

Can you please tell me why the headers of Exchange messages when using DAG indicate the primary hop through the replication network? Is it possible to configure the headers to only indicate the MAPI network, or is this the default behavior?

This is what I see in the message headers:

Received: from exchange01.domain.loc ([192.168.100.11]) by exchange01.domain.loc

([192.168.10.11]) with mapi id 15.02.1258.025; Mon, 25 Dec 2023 14:44:10

192.168.10.11 - address in the network "MAPI"

192.168.100.11 - address in the network "DAG Replication"

Exchange Server 2019, DAG with 2 nodes. 2 networks: 1 for Replication, 1 for MAPI.

3 answers

Your answer

Answer 1

Thameur-BOURBITA 36,261 Moderator

Hi @Step to IT

I think you should adjust the network configuration of your your DAG to ensure that The REPLICATION network is isolated from other networks (MAPI).

You should also check if you there is no DNS record for the IP used for replication network.

I invite you to follow the link below for more details about how you can isolate MAPI and replication network in DAG:

Exchange 2016 Database Availability Group

Please don't forget to accept helpful answer

Step to IT 125 Reputation points

2023-12-27T08:15:57.4866667+00:00

Hello!

The network is isolated, there are only 2 Exchange servers on this network, no other server has access to it. The network itself is used for traffic replication, there are no problems with this.

The problem is that the client connects to the server via MAPI and the server determines this address in the first hop, and then the rest of the headers proceed normally.

What is described in the attached instruction was done in the initial stages of preparing the DAG.

Maybe there is something additional that should be checked?
Thameur-BOURBITA 36,261 Reputation points Moderator

2023-12-27T11:29:39.0233333+00:00

Hi @Step to IT

In DAG network configuration, You can use powershell to disable replication traffic on MAPI network but we cannot disable MAPI traffic on replication network via Powershell. So the both network can be used for MAPI if they have a Gateway in network configuration.

The only way to ensure that network adapter is used for replication is to follow the configuration in the links below:

Replication network adapter configuration

MAPI network adapter configuration

Please don't forget to accept helpful answer
Step to IT 125 Reputation points

2023-12-28T10:27:11.5533333+00:00

I'm telling you, the network settings are done properly. The replication network is used only for replication (no gateway is specified, DNS is not specified, DNS registration is disabled, unnecessary parameters in the adapter "client for Microsoft networks" and "access to files and printers for Microsoft networks" are disabled).

Output of the Get-DatabaseAvailabilityGroupNetwork cmdlet relative to the replication network:

MapiAccessEnabled : False

ReplicationEnabled : True

Only replication traffic passes through this adapter. I don't understand why it appears in the message headers :(
Andy David - MVP 157.8K Reputation points MVP Volunteer Moderator

2023-12-28T15:04:17.0066667+00:00

I think that is because Exchange will simply pick whatever IP it sees on that server, not necessarily that its using the REPL network to actually transmit the message.
Step to IT 125 Reputation points

2024-01-02T15:24:49.6733333+00:00

Isn’t it possible to force him to use a specific adapter? And message headers have such entries on any DAG installation?
Thameur-BOURBITA 36,261 Reputation points Moderator

2024-01-04T11:27:19.3233333+00:00

@Step to IT

Unfortunately I think it's not possible.

Based on my experience one of my customer had the same behavior and finally after contacting microsoft support he decided to use one adapter for MAPI and Replication.

Please don't forget to accept helpful answer and close thits thread

Answer 2

Andy David - MVP 157.8K MVP Volunteer Moderator

There is very little reason typically to have a dedicated replication network. I haven't set that up since Exchange 2013. The preferred architecture is to have a single network for both REPL and MAPI, so I would consider removing REPL network:

https://techcommunity.microsoft.com/t5/exchange-team-blog/the-exchange-2016-preferred-architecture/ba-p/604024

Step to IT 125 Reputation points

2023-12-27T11:18:34.1833333+00:00

This option, of course, will be a solution to the issue. However, since the recommendation from Microsoft is to have two adapters, 1 for MAPI, 1 for DAG Replication, and everything works correctly in this mode, I want to solve the problem with headers and use best practices.

Answer 3

Andy David 701

Sorry, that is not the recommendation from Microsoft. The recommended configuration is listed in that doc I linked earlier.

Best practice is to use a single NIC. Much easier to setup and troubleshoot and will fix your issue. :)

https://techcommunity.microsoft.com/t5/exchange-team-blog/the-exchange-2016-preferred-architecture/ba-p/604024

DAG Network Design

The PA leverages a single, non-teamed network interface for both client connectivity and data replication. A single network interface is all that is needed because ultimately our goal is to achieve a standard recovery model regardless of the failure - whether a server failure occurs or a network failure occurs, the result is the same: a database copy is activated on another server within the DAG. This architectural change simplifies the network stack, and obviates the need to manually eliminate heartbeat cross-talk.

Step to IT 125 Reputation points

2023-12-28T10:38:50.2166667+00:00

https://learn.microsoft.com/en-us/EXCHANGE/high-availability/plan-ha?view=exchserver-2019

In previous versions of Exchange, we recommended at least two networks (one MAPI network and one Replication network) for DAGs. In Exchange 2016 and Exchange 2019, multiple networks are supported, but our recommendation depends on your physical network topology. If you have multiple physical networks between DAG members that are physically separate from one another, then using a separate MAPI and Replication network provides additional redundancy. If you have multiple networks that are partially physically separate but converge into a single physical network (for example, a single WAN link), then using a single network (preferably 10 gigabit Ethernet) for both MAPI and Replication traffic is recommended. This provides simplicity for the network and the network path.

Why not use redundancy if it can be useful in general? After all, everything necessary for this is available.
Andy David - MVP 157.8K Reputation points MVP Volunteer Moderator

2023-12-28T15:03:06.62+00:00

Well, the reasoning behind that if there as issue, the DAG will want to activate the databases on another server to mitigate. In other words, the additional overhead and complexity of the redundant networks does buy you anything when you can simply failover to another server. Thats the reasoning at least :)
Anonymous

2024-01-02T03:20:37.37+00:00

If Andy's answer is helpful, please consider "Accept Answer" and kindly upvote it.

Regards
Step to IT 125 Reputation points

2024-01-02T15:27:56.9566667+00:00

Thanks for your answer, maybe I'll try to disable the second adapter later and not use it now. However, this is a workaround, I would like to at least understand how to configure this if necessary...
Andy David - MVP 157.8K Reputation points MVP Volunteer Moderator

2024-01-04T12:45:18.9833333+00:00

Hi there, just to be clear, its not a workaround :)

Its the recommended way of configuring a DAG, one network. Good luck and let us know!

Share via

DAG Exchange messages headers

3 answers

Your answer