Monitor the health of analytics/notifications
As a user of asset-analytics and notifications, I need an easy way to know that PI Analysis and PI Notifications services are in good health and that event frames/notifications are being generated/sent as expected
In the PI System Health Tool, we are currently planning to detect health problems for AF, Notifications and Analytics in the 2018 R2 timeframe.
Do you have additional feedback / requirements about how you currently detect when Notifications have stopped working?
AdminJanelle Minich (Admin, OSIsoft) commented
Jon - we are looking at a way to enable the use case you describe - allowing users to modify AF without having to use PSE or PI Builder.
Tom LeBay is leading that work, he should be able to provide you with an update, and would probably enjoy hearing more about your use case.
This will fit into the larger Manageability Portal that we are just starting to work on, of which PI System Health will be a part.
Jon Croonenberghs commented
I have a customer who wants System Owners, Engineers and Managers who don't have access to PI System Explorer or PI Builder to be able to view the associated developed analytic for a mapped PI Tag with a PI Analytic Pointsource. Would this be an option for your PI System Heath Tool?
Alassane L. Seck commented
In response to Nebojsa Krstic, "I have a serious issue with the bug wher..."
I believe the known issue you are referencing is 117143. That issue is fixed in the latest version of Notifications.
I have a serious issue with the bug where the notification queue gets corrupted, Imagine we have thousands of assets having 10-15 analysis run or each and a dozen of notifications per element. Imagine what kind of a mess it becomes when the queue gets corrupted! Notifications go crazy/ resend... a mess in short. Which also makes it not reliable for surveillance as well. We can't trust it...
So if there is a patch or work around I would really appreciate it!
The 'solution' we currently use involves a Python script on a Scheduled Task, AF SDK, various counters mapped into PI points, and a dedicated analysis and notification that fires at the top of every hour. We look to see that the analysis counter and successful notifications emailed counter each increment and send email if not.
This is not as timely as we'd like. And it is a somewhat unique and obscure solution. I'd like to see something that mPI would pick up on, and a 'health' tag for each system (analysis and notifications) that could be monitored, put on a dashboard, or used to more easily trigger emails outside of Notifications.
Further, our "solution" simply tells us Analysis or Notifications are not working. Anything you can do to provide information about why it isn't working would also be great.