This AF kit will jumpstart better visibility into your PI Buffer Subsystems. The buffering system is a critical component for capturing continuous data from PI interfaces and from analytics engines. Buffers prevent gaps in data but often can suffer poor health that is unnoticed until users see gaps in data. Early notice can help you fix problems faster and find hidden root causes of issues.
I am providing here an AF template for PI Buffer Subsystem Health that reads PI Buffer Subsystem performance counters from Windows OS and sends simple alerts to keep you informed at the onset of problems. We archive the performance counters in PI using the PI Perfmon interface. We surface these tags then with an AF template. We capture event frames for each period of unhealthy performance.
In this template we have included analytics and an email notification message for the following situations.
Health is not "OK".
Out of Order Events/Sec rate is too high. (OOO events slow down your PI server and degrade response time)
Total Queued Events is too high. (Never good when buffer queue is filling and getting large)
Queue>Capacity is too low. (Always make sure you have disk space to cover a network outage)
In the kit too find a sample alert message format and tips to be more effective with email alerts. Subscribers to poor health alerts need an email when the alert starts and when it has cleared. This is important to avoid wasted effort by end users. We illustrate the steps to set that up properly here in this kit . We also illustrate best practices for building the body of health alert email messages. The kit is intended to be a starting point that illustrates techniques. You may find unique characteristics in your buffering systems that require you to add more analytics and alerts. Unzip and start by reading README file.
Hope this helps you get started. Please post your ideas and experiences.