I have 2 OPC tags set to grab timer preset and accumulated values on our default 1 sec driven scan class. These 2 OPC tags are updating each second as expected in the designer.
Now I have created a second expression tag that is set up to grab the difference between these 2 values and essentially give me a countdown of the process on the plant floor. This tag is set up on the same default 1 sec driven scan class.
The expression tag does not evaluate as expected. Sometimes this value doesn’t update for 20-30 seconds but it does not give an error. I have checked out the default scan class execution info on the gateway web interface, and it shows that the default scan class is executing each second without delay.
The issue is intermittent, but seems to start happening after the gateway has been running for a while. We had the issue on our main ignition server and recently migrated the gateway to another machine. The issue went away for about 3 weeks, but is present again so I would imagine it’s something caused by our configuration.
One thing that we’ve noticed that seems to coincide with the scan class evaluation issues we’re having is that historical data is missing from trends. We have had to set the “validate scan class executions” parameter to false on all of our easy charts to mitigate this, but I would rather be able to find the root cause of the issue than use workarounds.
Scan class timing is a best-effort affair – problems in one or some tags (say, a query tag that runs too complex a request) that delay execution will impact other tags. Look for long query warnings in your logs, and consider setting up a scan class just for these tags.
So I have been watching the logs under the “Tags.Provider.Execution.Scanclass” heading and keeping an eye out for errors.
I have watched the logs populate at the trace level for a few minutes and can see the scan classes executing without issue with one exception. At one point, we had 11 instances of the following message at the warning level:
Scanclass 28Feb2018 12:34:10 Received a lease expiration that was older than the previous, will not apply - Sat Aug 16 23:12:55 PST 292278994 (9223372036854775807)
I’m not sure if this tells me anything related to my issue.
So, up next, I have created a new scan class “Expression 1 Second” with the same parameters as the default scan class(Direct, 1,000ms, Subscribed). The tag continues to evaluate every 20-30 seconds and no sooner.
I appreciate the help in advance. Are there any other recommendations?
After 7.8 to 7.9 update we experience the same issue. We are using expression tags as scanclass driving tags, so the performance is very important. I see, there is no feedback from IA about the problem. Maybe anyone has figured out any solution?
Background information:
We have tested absolutely independent direct scanclass with 1sec execution time for expression tags. No way it updates faster than 30-40 seconds. The Status->Tags->Scanclasses information page does NOT show any performance issues.
Seems, that the problem is not version related. After things got worse and worse we installed 7.8.5 back and…got the same problem. After restore, first time the scanclasses are executed regularly with no delays. Then, after a hour or so, things started to get slower and slower.
Current situation:
Installed 7.9.8 back, revised all the scanclasses, some of them had “READ” subscription mode. Put “Subscribed” instead. Also exported all the scanclasses to CSV and imported back. For 1.5hours already we have had stable work, SC duration time is low, no problems so far.