Mail Queues

Evidence of a problem with the technologies that support Exchange Server 2003 usually takes the form of mail not getting sent or delivered. You can think of Exchange Server 2003 as having two mail queues, one for messages to be delivered to users in the organization and another for messages from users in the organization that are to be sent externally. In actuality, these are two protocol queues (X.400 and SMTP). The real number of queues is dynamic, one created for each outbound domain; however, when you are new to Exchange it can be helpful to think of two single queues. When diagnosing problems, be certain to compare the size of the incoming queue to that of the outgoing queue. The sizes of each might hint to the location of the problem.

Buildup of Messages in the Local Delivery Queue

If the number of messages in the local delivery queue is beginning to build up, it suggests that there might be a problem in Exchange Server 2003 receiving information from Active Directory or forwarding messages for local delivery. A buildup of messages in the local delivery queue should suggest to you that there is not a network infrastructure problem, as messages are clearly arriving at the server from elsewhere on the network and the Internet. There are some exceptions to this rule, such as the failure of an external DNS server that would allow messages to come to the server but might stop messages from being sent.

Buildup of Messages in the Outbound Queues

Message buildup in the outbound queues indicate that there are problems with the network or problems with the servers that should be processing the mail. When there are problems with the outbound queues, remember to look at what is happening with the local delivery queue. If the local delivery queue is empty, or close to it, and nothing seems to be entering it, mail is not being received or sent. This could mean that a smart host has failed or that the Exchange server has lost network connectivity entirely. If the local delivery queue appears normal but the outbound queues do not appear to be processing messages, look toward problems with how the external mail is being handled, such as whether the external DNS server used for resolving the names of remote Simple Mail Transfer Protocol (SMTP) hosts has failed.

The tool that you should use to diagnose queue problems is the Queue Viewer, as shown in Figure 10.1. The Queue Viewer is located in the ESM utility. To find it, navigate under the Servers node to the Queues node. In the Queue Viewer, you are able to see the following:

• The number of messages in each queue

• The total size in KB of messages in the queue

• When the oldest message was submitted

• When the next connection will be retried

Figure 10.1. The Queue Viewer allows you to view statistics on each of the Exchange Server 2003 mail queues.

Perhaps the most useful of these statistics is the one that lists when the oldest message in the queue was submitted. This gives you a good idea as to how long it has been since the queue functioned correctly.

Performance Counters

The MMC Performance snap-in can be used to monitor a range of Exchange Server 2003 or Windows Server 2003 functions. The following performance counters are useful for monitoring mail queues:

• SMTP Server\Local Queue Length Messages destined for local delivery. The value should be less than 1,000.

• SMTP Server\Remote Queue Length Messages destined for remote delivery. The value should be less than 1,000.

• MSExchangeIS Mailbox\Send Queue Size Number of messages in a particular mailbox store send queue. The value should be less than 500.

• MSExchangeIS Mailbox\Receive Queue Size Number of messages in a particular mailbox store receive queue. The value should be less than 500.

• MSExchangeIS Public\Send Queue Size Number of messages in a public folder's send queue. The value should be less than 500.

• MSExchangeIS Public\Receive Queue Size Number of messages in a public folder's receive queue. The value should be less than 500.

You should generate a baseline set of figures for each of these counters during normal server operation so that you are later able to quickly diagnose any problems that might have arisen.