Additems failed in OPC DA Server
One of customer system keep generating "AddItems failed in 800xA OPC DA server", "AddItems timed out in 800xA OPC DA Server" and "AddItems completed in 800xA OPC DA Server after a previous timeout" in random Servers and Clients node. The system is only have Advant master connect plus surrogate for PI. Those message "Additems xxxx" is mostly on Network level in this case Network 51 in control structure, there are some on object that also failed to Additems, but the object point to library object like PID Object (MUNIT properties) or AI object and not to particular tag number. Event details not showing any clue on the event generated.
This repeating event are somehow increase one controller load to 100%.
Any experience on similar case and how to troubleshooting this issue?
Thank you and highly appreciate for your reply
A bit difficult to comment on without having any additional information.
If you have circled the warnings to occur only when the PI client is connected (which is a pure guess from my side) the likely reason is that PI add too many items too fast and breaches the performance envelope of the relatively speaking, weak MB300 OPC server. The warnings you see only points to the problematic control network, not which client that is responsible for the overload.
- A highly loaded controller will significantly reduce the maximum achievable throughput. Stay below 80% and use SC520 carrier boards for CS513 (in AC450) and CI547 cards (in AC410) to offload network tasks from the main processor.
- Having the RTA Board Configuration tool open during problematic periods will likely bring important system messages to our investigation. A weak controller may be spotted since the "overflow" messages will contain node numbers in hex, to the far right, eg "DCXA1280 24 5 ... 0x330A..." indicates controller 330A = 51,10 is choked and cannot accept more subscription requests.
- Make sure the PU410 upper side connection with the AC400 Connectivity Server is running at 100 FULL duplex. The PU410 box cannot auto negotiate and any attempt to do so should result in the host's NIC falling back to 100 HALF and the ensuing duplex asymmetry will cause communication losses and a sharp drop in performance.
- Over the years I have probably posted "kilometers" here on AKS on the best practices with this OPC server for MB300 (be extremely cautious with writing, write only to MB300 DAT, never "scan", subscribe!, do not subscribe non-cyclic properties the controller cannot "pump" efficiently, only subscribe at 9 seconds update rate, etc, ...). Search and you should find more rich answers.
- AKS long time member Rob Lyon have several times promoted using PI's OPC HDA interface (read from history logs instead of directly from the controller).
I have attached a copy of the System Health Check manual, it has a chapter dedicated for 800xA for Advant Master, I suggest you have the system screened from top to bottom. I can assist with deciphering the results you can post here or to firstname.lastname@example.org. Your regional ABB support team should also be able to assist with the analysis or escalation of support cases.
Data Handling - staggered group activation - must be ON
Advanced Options - Enable Mass Tag Adding - must be OFF
These two settings will slow down the rate at which tags are added to the OPC server. I currently pull 2941 tags off MB300 using PI.
As Stefan has indicated - MB300 will suffer if you over subscribe or get the timing wrong. Our IT dept literally brought the plant to a standstill - now i look after PI.