The first SDK Yield thread error message is expected as a result of the win32 exception. According to some internal documentation:
“The SDK Yield thread is only responsible for processing health checks received from the PI server and pibufss heartbeats. The thread is usually stopped with other SDK threads when the process associated with the thread shuts down. The reason for the error may be due to an internal synchronization lock being held longer than the timeout for stopping the thread. Asked customer to check if the PI server is responsive and if there are other outstanding error messages, and was told PI server is responsive and there are no other error message. Since everything looks fine, no need to worry too much about this message.”
We need to find the root cause of the Win32 exception. The related error message is too generic. Do you see any error message on the PI Message log from the PI Data Archive and from the client? What was changed since the problem has started on the web server? Do you know which line is throwing the exception?
As far as we know, nothing has changed in terms of PI-SDK version or hardware. Forgot to mention that the issue is intermittent. The site will run for days and then completely stop responding for hours. This behavior coincides with the "SDK Yield thread" and JIT debugger error message. It happens at all different times of day. To recover, we have to recycle the application pool in IIS. We'll look at logs. Unfortunately it's intermittent so catching it may be a challenge.
What's the best way to get in touch with PI tech support regarding this issue? We found some info in the logs but so not want to post customer info, IPs, etc. on the message boards.