Administrators who use Exchange 5.0 over WAN links (e.g., the Internet) need to be mindful that there are Internet
traffic conditions that can cause problems not found in LANs. One of the most persistent problems is slow remote links or links that are being hindered by packet loss or bad switching. Exchange 5.0 is not tuned by default to be more aggressive with re-establishing connections under poor network conditions. As a result, when network "weather" is bad, Exchange 5.0's MTA (Message Transfer Agent) may repeatedly log errors with the event ID 9318.
The exact error may vary, but it will look something like this:
Event ID: 9318
Description: An RPC communications error occurred. Unable to bind over RPC. The locality table (LTAB) index is 76.
To compensate for persistent bad WAN links, try this:
- In the Message Transfer Agent's Site Configuration Properties window, select the Messaging Defaults tab.
- Set the Checkpoint Size (K) box to 15.
- Set the Recovery Timeout (sec) box to 90.
- Set the Window Size box to 3.
- Click OK and close the MTA Site Configuration Properties window.
This workaround sets several properties of the way the MTA handles WAN links -- most importantly, the Recovery Timeout setting, which is how long the MTA waits after an error before giving up on trying to recover a dead link.
The Checkpoint Size is how much data is sent before the MTA tries to establish a checkpoint (i.e,. where to continue resending data from in the event of a failure). This value can be set higher if you have fewer errors (by default it is 30).
The Window Size is the number of checkpoint failures that must take place before MTA aborts the data transfer entirely.
Serdar Yegulalp is the editor of the Windows 2000 Power Users Newsletter.