I am connecting Ignition (v8.1.44) to an opc-ua server (OPC CLIENT -> OPC Connections -> 3rd party opc server) and I am noticing that there is a timeout every two minutes in the log, I also see the tags go to Uncertain_LastKnownValue and then back again around the same time.
I enabled debug on the log and got more detail on the error
based on the post below from the opc-ua forum, the error seems to have to do with the TimeoutHint value.
This post on the ignition forum talks about the TimeoutHint but its the opc-ua server not responding properly. (if I understand properly)
UaTcpStackClient: Received PublishResponse for unknown requestHandle - #8 by chi)
Here is my wireshark packet trace, you can see the connection (858), then the retransmission with a push (860). (IP's removed). This occurs every two minutes on the button, it seems to align with the Request Timeout (120,000 ms)
Lastly, when I connect with UA Expert I have to set the Stack.TcpConnection_DefaultChunkSize = 131072. Not sure if that is helpful but maybe a setting I need to change to establish stable connection.
I know this is a lot and I haven't called support yet but I figured I would run it by you first. Does anyone have any suggestions or tips, thanks!
If you can get a multiple minute long Wireshark capture that includes creation of the subscriptions and the timeouts, and is taken without security enabled, it would be useful. Basically, disable the OPC UA connection in Ignition, start the capture, enable the connection, wait however long it takes, then stop the capture. Then upload the capture file and not a screenshot.
Beyond that you're probably going to need support from whoever made that server because this probably means a bug in that server.
1 Like
Ok, I was able to remove encryption and do a wireshark capture. We are seeing a UA Secure Conversation Message: DeleteSubscriptionRequest every two minutes that coincides with the disconnects.
I believe that is standard part of the UA protocol, any suggestions on next steps? I can call support and submit the packet capture too. thanks!
I've sent you a DM with a dropbox link you can upload the capture to.
If you have Ignition logs from the time of the capture those are useful as well.
1 Like
Okay, it's exactly like I expected.
This Server is failing to send PublishResponses in a timely manner, it's triggering the Subscription watchdog timer on the Ignition client side.
The Subscriptions get created such that a keep alive response is expected every 60s in the absence of a PublishResponse containing actual data changes. The Ignition client waits 2x this period (120s) and then concludes something is broken with this Subscription on the server, deletes it, and then eventually creates it again.
You need the Server vendor involved, because it seems this Server implementation is broken.
I also noticed some ServiceFaults with Bad_Timeout that are responses to PublishRequests that were sent with a 180,000ms timeout hint, but the fault is sent back after only ~30 seconds. Again seemingly something broken in this Server.
1 Like
Thanks again! I appreciate the help and I have contacted the vendor.
