MTA Slow on LAN when all WAN Connectors are fine...

[jzakora]

I have an Exchange 5.5 environment in North America that is having issues right now.

The environment:

Essentially we are hub and spoke with X400 connectors. We have a "hub site" containing two machines with IMC's to the internet relay. We have 17 Exchange "Mailbox Sites", including one on the same physical LAN as the Hub Site, that are connected to the hub Exchange Site via an X.400 connector on each Hub machine. These connectors are always on, no limitations.

The Problem:

The Mailbox Site on the same LAN as our Hub Site is having the issue. I've been noticing the problem every day for a month now, since they downgraded our WAN a bit. I watch all the queues in North America flow all morning, a bit slowly because of the WAN downgrade, but this one just keeps climbing and it's the only one on the LAN which I would expect to flow fine regardless of WAN connectivity. By noon it's got 1500 mail in the MTA queue on both hub servers trying to get to the server in question. Additionally there will be 1000 mail in the queue trying to get up to the hub site for delivery to the other sites.

Today I got new information. We had a backup kickoff around 2pm on the hub servers. Immediately the queues were flushed down to the mailbox site and on that server up to the hub site. Just like that. One minute there are 4000 messages total in all the queues, the next there are none.

Is there some command that the backup agent is issuing that forces a flush of the queues?

Does anyone have any other ideas as to what might be happening on my system here?

 

[jzakora]

Come on MVP's. None of you have an idea on this one? Please, I need to fix this, I'm again watching my queues climb and my mail delivery times slow down accordingly.

Any help at all is appreciated. Thanks in advance.

 

[DirtDawg]

For the messages queued up what are the details of them being delayed. [Host unreachable] or something like that. you might also check to the connections accepted by the server by doing a telnet port 25. what backup software are you using?

I will try to help you if I can

 

[DirtDawg]

Just a another quick thought what are the retry intervals set at on the servers

 

[jzakora]

Since posting this request I've been able to get some kind of hold on the queues. First we made some WAN changes to lighten up the load on the primary links. This helped a bit but I believe the biggest help was from some registry tweaks HKLM\System\CurrentControlSet\Servies\MSExchangeMTA\Parameters
TCP/IP control blocks=420 (was 20)
TP4 Control blocks = 420 (was 20)

This has helped the queues quite a bit.

The delay is not evident. They flow, just incredibly slowly, well better now after the reg tweak.

Retry interval I have to look up. I'll reply again.