Problem with PU410 unit
I have a problem with one of our redundant connectivity servers PU410-unit. We are running version 6.0.3-3.
The load is fluctuating much more on one of the PU410-unit than the other in the same redundant connectivity server. Sometimes i even get the temporary high load in OPC DA alarm and there is a red cross over the process objects.
Is there some log file or anything that can give us a hint of the problem ?
Voted best answer
Many wise words have already been mentioned here by others.
Let me recapitulate a few cornerstones:
- Do NOT write to any other object type than MB300 DAT
- Write with low intensity (the MB300 design from the 1980-ies was cut for a handful of operator stations with humans at the keyboards, not robots with "auto-fire" capabilities)
- Do not have clients "scanning" for data on their own interval, not even 9000ms, always subscribe at 9000, 3000 or 1000ms
- Do not subscribe for properties lacking generic cyclic support by the controller
- MB300 AI:s RANGE_MAX, RANGE_MIN, ALARM_DELAY_PERIOD are NOT OK
- MB300 AI:s VALUE, STATUS, NO_OF_DEC are OK.
More information about what object type and property having generic cyclic support is listed in the 800xA for Advant Master Configuration User's Guide.
- Do not have occasionally "reading" applications, restrict clients to have active items in active grups. Then read from cache or use Advise() to receive asynchronous updates whenever data point or quality changes. See attached PDF document about OPC function calls.
Be observant on System Messages output in the RTA Board Configuration tool connected to the PU410 or Control Builder A's OnLine Builder when connected to the controller. Post them here (entire row with all hexadecimal values) and I will comment.
Use ANPER Channel Utilization Analyse to find bottlenecks in PU410 or controller. Post results here and I will comment.
Please find some more hints in the attached System Health Check document, there are two main sections applicable for your issue: OPC DA Connector health" and the entire 800xA for Advant Master chapter.
*** UPDATE: uploading attachments to AKS seem disturbed right now. The files become cropped (should be 5.6MB and 277KB respectively)
i may not answer your question directly, but here are some helpful bits....
MB300 comms recovery using the old RTA board or the PU410 is not particularly fast. Even if you have two good CS and PU410's and disconnect one, the switch over to the remaining connectivity takes a bit of time and usually results in red crosses.
From my experience, we had intermittent red crosses for a long time and could never find anything wrong, until i started looking at the media convertors on each controller, especially the collision lights. Then i knew i had a network problem and not a PU410 issue. I changed my hubs out for a simple un-managed netgear switch, as well as changed out all the UTP cables to new CAT5 cables.
The collision lights stopped after that, and no more red crosses.
You may actually have a high load because one PU410 is carrying the load, even though everything appears to be working. See attached image..
have you added the RTA Load object to your RTA board - it will trend the load
Use 9sec scan wherever possible.
Check also if any Property Transfer aspect, which is reading from Advant controller, is set for other interval than 1 / 3 / 9 sec.
I can’t find any system messages from PU410 or the AC450 controllers ONB.
I will post the result from Anper first for PU410 and then the AC450 Controllers.
It would be great if you can point in the right direction of the problem.
The PU410 channel load does not look too bad
There is some activity on DCSCN2, the channel used for >9 second cyclic subscriptions and for properties the controller cannot send cyclically forcing the OPC server to repeatedly ask for them which drains resources and reduce performence.
If possible, identify the subscribers behind those properties and correct them to gain some performance
The document I post here gives info on how to use applog to study subscribers connecting an OPC server in 800xA and also how to view what each client is having added to the servers. The "Statistics" operations are safe to use and can tell you a lot of an OPC server and OPC client's behavior in 800xA.
Also, try to find System Messages at the time of the problem and have the DA Connector and MB300 Adapter Statistics run at that time.