The Exchange Server Troubleshooter - 08 Sep 2000

Can I alter the default text in Exchange nondelivery report (NDR) messages?

Users had hoped that Microsoft would offer this feature in Exchange 2000 Server, but it didn't materialize. Neither does Microsoft offer a supported way for users to change the default text. At present, you're stuck with the default text. One cool thing you might do in Exchange 2000 is to write some event sink code to send a custom message when Exchange processes an NDR, but writing this code might be more trouble than it's worth to you. (Thanks to reader Geraldine Cardwell for this question.)

To use the disk space on my server, I want to move the dsadata and mtadata folders to the same drive as the mdbdata folder. Is a simple method available to move these files?

Run the Performance Optimizer. The locations of these files appear on the fourth and fifth wizard pages. You can move the files just by specifying a different destination. However, don't try to move the files manually; each transaction log file contains a signature that points to its physical location on disk, and moving the files to a new location renders this signature invalid—thus making the log files unusable. Moving the files with the optimizer avoids the problem. To protect your system from the effects of disk failure, always run a full online backup immediately after you make the move. (Thanks to reader Tim Graham for this question.)

In your July 2000 column, you suggested using Exmerge to move users in one site to another. Can you also use the Move Server Wizard?

Astute reader (and Exchange Most Valuable Professional—MVP) David Sapery caught me here. Let's say you have two servers in two sites (e.g., A and B). You need to move the users from a server in A to a server in B. One way to move the users is with Exmerge, but you can also use the Move Server Wizard to move the server in site A to site B, then move the users with the Exchange Administrator Tools, Move Mailbox command. This method is easier than using Exmerge, and the Move Server Wizard is smart enough to update the internally stored X.400 address for each message so that replying to a message generates the correct return address.

Exchange 2000 lets me use multiple Information Store (IS) databases on one server. How many IS databases can I have?

Some confusion exists about the number of databases that one server can host. Exchange 2000 introduces the concept of storage groups (SGs), or collections of databases. Each SG can contain one or more databases, and you can have more than one SG on one server. At the 1999 Microsoft Exchange Conference (MEC), Microsoft announced that beta 3 of Exchange 2000 would allow 5 SGs with 16 databases each—a total of 80 databases per server. Later releases of Exchange 2000 lowered the total to 4 SGs, each with 6 databases per server—a total of 24 databases per box. However, the practical limit is 20 databases: 4 SGs with 5 databases each. Why? Because in Exchange 2000, you can perform maintenance work (e.g., restores or running Eseutil) on one database in an SG without affecting the others, provided you have a free database slot. Thus, if you have 1 SG with 6 databases and you need to restore a database, you'll have to unmount one of the other databases to free up a slot. Limiting yourself to 5 databases per storage group preserves that slot. And if you couple that number with the limit of 4 SGs per server, you arrive at the final answer of 20 databases per server.

I want to export our users' contact folders into one big server-side contacts folder. Can I perform this export automatically?

Recent Exmerge versions let you perform this exact operation; you can even specify which folders to export and merge. As a reminder, you can get Exmerge from Microsoft Product Support Services (PSS); if your organization has a premium support contract with Microsoft, your technical account manager can also get Exmerge for you.

We're using the Windows 2000 version of the Active Directory Connector (ADC) to replicate our Exchange Server 5.5 data to Active Directory (AD). However, the default replication interval the ADC uses when we choose Always is 5 seconds, which is too short. Can we adjust this interval?

You can adjust the replication interval for the ADC by setting a particular time when you want replication to occur. Another option is to set the interval to Always, then increase the interval to whatever value you think is appropriate. You make this adjustment by adding a new REG_DWORD value named Sync Sleep Delay (secs) to the HKEY_LOCAL_MACHINESYSTEMCurrentControlSet ServicesMSADCParameters Registry key. Set the new DWORD value to the number of seconds at which you want replication to occur.

I'm running two Exchange servers, one of which is accepting SMTP mail for the other. However, mail dequeuing isn't working properly. How can I solve this problem?

Dequeuing is a simple idea that is sometimes difficult to get working. The basic idea is that one server (e.g., A) accepts SMTP mail for another (e.g., B). At some interval, usually controlled by B, B connects to A and requests a transfer of any queued mail for B that A is holding. The best way to implement this process is by using the SMTP extended turn (ETRN) command, which Microsoft designed expressly to let one server dequeue its mail from another. The best source I can think of for general-purpose dequeuing help is Simpler-Webb's SMTP FAQ. (Thanks to reader Steven Barnard for this question.)

How is disaster recovery in Exchange 2000 different from that in Exchange Server 5.5?

This question deserves an entire article because many differences are subtle. However, you'll notice some major differences right away. First, in Exchange 2000 you don't back up the directory as part of Exchange anymore, because AD is part of Win2K. You still must back AD up, of course, but to restore a downed server, you restore the OS and AD before attempting an Exchange restore. The presence of multiple databases on a single server can complicate your planning too, because in many cases you can restore a damaged database without affecting other operations concurrently running on the server. You also must get used to the Win2K Backup utility, which is different from its predecessor. One other potential gotch Exchange 2000 depends on the IIS 5.0 engine—part of the Win2K OS—to handle IMAP, POP3, HTTP, and Network News Transfer Protocol (NNTP) traffic. IIS keeps its configuration in the metabase, so make sure that you include the metabase data in your regular backup rotation.

What does event ID 202 mean?

Event ID 202 is produced by post-Service Pack 3 (SP3) versions of the Exchange Server 5.5 store process. The event is informational, but because you don't see it before you upgrade to SP3, its sudden appearance might confound you. The event is telling you that the store has built a cache for a particular type of database index called a space tree; Microsoft includes the event in the log so that PSS can better support customers with very large databases. You can safely ignore it. (Thanks to reader Tshisekedi Mbuyi for this question.)

Can users on an Exchange Server 5.5/Windows NT 4.0 server use an Exchange 2000 server for their public folder server?

Sure! In fact, you can use Exchange 2000 with Exchange Server 5.5 several ways. Exchange 2000 can provide several services to mailboxes homed on Exchange 5.5 servers:

(Thanks to reader Joel Baum for this question.)

We're bringing up a new Exchange Server 5.5 server in a remote location through a Site connector as part of our existing organization. To save time, we're thinking about configuring the server at our corporate office first, then shipping it to the remote location. Will this approach work?

As long as the freight company doesn't mangle your server, this approach will work fine. Set up the server in your corporate office, install Exchange, add a Site connector, and let the directory replicate. When you're confident that everything is working correctly, pull the plug on the server, box it up, and ship it off. When it arrives at its destination, unbox it, change its network settings, and verify that replication traffic is flowing normally across the Site connector. (Thanks to reader Mark Senia for this question.)