Distribution Agent reports the transactional replication latency incorrectly in SQL Server 2008, SQL Server 2008 R2, or SQL Server 2012

When you experience this problem, alerts that you have configured to detect the latency of transactional replications are invalid. For example, you set up an alert that notifies you when the latency of replication is larger than 30,000 milliseconds. However, the alert frequently reports periods of latency when the subscription is actually synchronized.

To work around this problem, follow those steps:

1. Create a second identical alert that detects a smaller latency value (for example, 20,000 milliseconds).
2. Configure the second alert to execute a job that inserts a tracer token into the publication. To do this, use the following statement when the condition is true:
   EXEC sys.sp_posttracertoken @publication = '<SubscriberDB>'

When the second alert is triggered by an incorrect latency value, the execution of this procedure will clear the problem. If the latency value is correct, the latency will continue to increase past 30,000 milliseconds. Therefore, the first alert is triggered correctly, and you receive notification of the correct latency value.