One of our Exchange Server 2003 servers recently began throwing off numerous instances of event ID 1159 from the Information Store. Stopping and restarting the Information Store several times eventually stopped the event from being generated—but what caused it?
The problem happened because the Information Store couldn't advance the checkpoint that marks which transaction log file is currently in use. During a backup, the Extensible Storage Engine (ESE) backup API advances the checkpoint to reflect which log files' transactions have been committed successfully. Sometimes, the Information Store can't advance the checkpoint—something that happens usually because an object in a transaction log can't be flushed. The checkpoint depth (which you can monitor by using Performance Monitor's Log Generation Checkpoint Objects counter on individual database instances) should remain well below 100. Microsoft recommends that you take troubleshooting steps if the checkpoint depth remains above 100 for any length of time.
The most common cause of the problem is a backup-application failure. When an ESE-aware backup program starts a backup, the Information Store service expects the program to finish by making an appropriate API call. Until that API call is made, the Information Store thinks the backup is ongoing. When you restarted the Information Store, you effectively removed its knowledge of the previous backup, and it could start soft recovery of the backlogged log files.
What you should ask yourself now (presuming this explanation fits your circumstances) is how long your backup software has been failing. Event ID 1159 won't appear until you exceed 1008 log files; most organizations take a significant length of time to reach that number of log files. Take a look at the Microsoft article "Information Store databases do not mount and you receive an 'Operation terminated with error -614' error message when you start a computer that is running Exchange Server 2003" (http://support.Microsoft.com/?kbid=836611) for more details about the problem and how to get the appropriate hotfix. For more information about how ESE works and some troubleshooting suggestions (albeit for Exchange 2000 Server), see "7 Daily Checks to Keep Exchange 2000 Running Smoothly," October 2002, InstantDoc ID 26185 and "The ESE Backup Process: An Inside Look," May 2002, InstantDoc ID 25350.
