XADM: Master Merge May Appear to Be Stuck at "Paused - Flushing"

During a master merge, the master merge may appear to be stuck at "Paused - Flushing," which indicates that the merge has been paused in the middle of the process. The following error message may be logged the Application event log in Event Viewer: However, there is no indication of the type of error (for example, whether the error is low disk space or something else), nor is there an indication of which disk may be low on space.

If you use Microsoft Windows 2000 System Monitor (Perfmon.exe), open the Microsoft Gatherer object, and then look at Reason to back off counter, the counter value is 6.

To resolve this problem, obtain the latest service pack for Microsoft Exchange 2000 Server. For additional information, click the following article number to view the article in the Microsoft Knowledge Base:

After you apply Microsoft Exchange 2000 Server, Service Pack 1 (SP1), the following event description in included in the error message that is logged in the Application event log for the event that is described in the "Symptoms" section of this article:

Microsoft has confirmed that this is a problem in Microsoft Exchange 2000 Server. This problem was first corrected in Microsoft Exchange 2000 Server Service Pack 1.

A master merge occurs when you use the content indexing option in Exchange 2000 Server and enter an update interval. Every time that the MSSearch engine updates the index (or re-index), the MSSearch engine does a master merge of the index table.

If the pause occurs because of low resources, an event is logged in the Application event log as described in the "Resolution" section of this article.

If low resources are detected during a crawl, the Gatherer enters a "back-off" state, and Search pauses. The reason that the service is paused (either because of the preceding conditions or manually by the administrator) is displayed in the following locations:

• The Microsoft Management Console (MMC)
• The status view in the Content Sources folder

If the gatherer is in back-off state, a status of "paused � back off" is displayed.

You can find the exact condition that caused this status by monitoring the Perfmon counter in the Microsoft Gatherer object under Reason to back off. If this counter is not zero, the Gatherer is in the back-off state. The error codes corresponding to values are:

• 0 - Up and running
• 1 - High system input/output (I/O) traffic
• 2 - High notifications rate
• 3 - Delayed recovery in progress
• 4 - Back off because of user activity (manually paused)
• 5 - Battery low
• 6 - Memory low
• 99 - Back off for internal reason (forced by search itself, most likely an unknown error condition)