|
EMAIL System Performance Monitoring |
|
||||
|
|
The exchange alert system consists of the following components:
The mail injecting process sends a series of 6 messages at five minute intervals. These messages are sent to each of our inbound mail servers and to an external mail server at CIT. The recipients for these messages are on our Exchange Mail Server as well as our lists server. As each message is sent a log entry is created in the logging database. The log contains a unique ID for each message sent as well as the sent time of the message.
The IMAP receiving process then attempts to download and read each message. As the messages are received they are looked up in the logging database. When the lookup is successful the received time as well as the routing information for the message is saved in the database.
When this process is complete the log is analyzed to determine if the messages were received within an appropriate period of time. If not an alert is generated and sent to both CFS and SMS pagers. If the messages are subsequently received and no additional alerts are generated then an Alert Resolved message is generated and sent to CFS and SMS pagers. The inbound message queues are monitored to ensure that there is not a backlog of inbound messages. If the number of messages in an inbound queue exceeds 45 messages then an alert is generated. If the number of messages subsequently goes below this threshold then an Alert Resolved Message will be sent. Otherwise additional alert messages are generated. There is a “fail safe” process built into the system that will send additional alerts should either the message delivery time or the queue lengths exceed 10 minutes or 90 messages.
A heartbeat message containing statistics about the previous 24 hours of operation is sent out 3 times per day. At 6:00am, 2:00PM and 10PM this verifies the functionality of the alert system itself as well as alerting CFS to performance trends or performance slowdowns
