I have a bit of an issue with and OPC interface seemingly using up to 99% of the CPU loading on the interface node. This is linked to the scans skipped value for that interface rising, and the heartbeat value for the interface taking several minutes to complete its 15 second sawtooth profile. I've attached a file showing a PI trend of the interface statuses, along with a PI tag which is used as the watchdog. The watchdog timer seems to time up correctly, but when I compare it to the actual value on our control system, then it is completely different (1040 on the control system, 2246 in PI - the watchdog counts up every second then resets every hour, so it would seem that PI is lagging behind the actual value.
I've also attached a screenshot of the CPU loading and the error message generated in the pipc logfile. The bit about the evmexceptions queue doesn't seem to me to be a good thing.
Can anyone give me some pointers on what might be causing the issue? Can someone explain in simple terms what the scan skipped value actually means; is the number of tag updates missed, or is it the number of scans missed by an advise group, or is it something else. Also why would the IORate value increase when the interface seems to be slowing down?
Just for clarification, this interface is 1 of 3 we have running connected to the same OPC server, all running on the same node. We have an interface for digital tags, analogue tags, and string tags. The advise group size is set to 400 tags in the analogue interface, and 200 for the digital and string tags. There are roughly 1500 digital tags, 2500 analogue tags, and 100 string tags. The interface process for the
analogue tags is the one which seems to be using 99% of the CPU usage (I found this out by killing that process and seeing which interface stopped in the ICU).
OPCint version 184.108.40.206, UNiINt version 220.127.116.11
Any help much appreciated.