First of all, can you please enhance you logs by timestamps. Without having the time associated with each log entry, the log is nearly useless.

Second, the logs show communication problems with the server, so I do not see why these should always be looked at by our side. For example, the log entry clearly says that the server is not responding to keep alive requests (that can be network or server issue). I have no reason to think that these error log entries are wrong, until shown otherwise.

Best regards

It is possible that the difference between message sizes (estimated # of tags) I saw was not real - unfortunately, as I wrote, precisely at the point where needed most, Wireshark failed to show the true content of the communication.

Thank you for the Wireshark logs. I do not have a conclusive answer for you, but here are my findings:

A) The network appears to be not just slow, but in a "bad shape". For example, in the opclabs log, there seems to be no truly delivered OPC-UA messages since time around 106.26: Only TCP retransmissions after 60 seconds each (times cca 166.97, then 226.97, 286.97, 346.97, 406.97, 468.98 ... etc.). The kepware log shows problems too, although not as such bad scale. Longer messages in particular seem to be quite problematic to process or deliver.

B ) In the opclabs log, the requested publishing interval for the main subscription is 500 milliseconds (I suppose this is because you use a sampling rate of 1000 milliseconds), whereas in the kepware log, the publishing interval is 5000 milliseconds. In order to make a fair comparison with Kepware, you should use a publishing interval of 10000 milliseconds in your QuickOPC code. The publishing interval forms a basis for computation of keep-alive interval, and has therefore influence on communication problem detection.

C) For some reason, Wireshark does not show me the decoded contents of the CreateMonitoredItemsRequest message in either log, therefore I cannot easily compare them in detail. But I can roughly judge from the length of the message, which is 15410 bytes in opclabs log (packet #7217) but is 1370 bytes in the kepware log (packet #29501), that the number of items used from the Kepware client was much lower than that used from QuickOPC. Correct me if I am wrong. If you want to see results comparable with Kepware, you need to subscribe to the same (number of) items. My estimate is at the opclabs request contained about 200 monitored items, while the kepware request around 15.

D) QuickOPC is correctly flagging the session as halted because there truly is no communication from the server for extended period of time (which can be for the reasons described under A, B, C), but if needed, the time interval for such detection can be increased, making the client to be more "patient" when something in the communication goes wrong.

My recommendations are - please try them in this order, first 1), then add 2), then add 3):

1. Increase the FastestKeepAliveInterval from 5000 to e.g. 30000 milliseconds, and the FastestLifetimeInterval from 15000 to e.g. 90000 milliseconds (this to address the finding under "D"). For client objects that have EasyUAClient.Isolated equal to 'False', these settings are accessible through EasyUAClientConfiguration.AdaptableParameters.SubscriptionParameters. For client objects that have EasyUAClient.Isolated equal to 'True', these settings are accessible though EasyUAClient.IsolatedParameters.SubscriptionParameters.

2. Increase the sampling rate of your monitored items from 1000 milliseconds to 10000 milliseconds.(this to address the finding under "B")

3. Call SubscribeMultipleXXXX in "batches" with reasonable number of items (such as the number you have used form Kepware client) - that is, not one item at a time, but also not 200 items at a time. Start with e.g. 20. (this to address the findings under "A" and "C")

Best regards

Dear Sir,

thank you. I apologize but I won't be able to look at the logs sooner than next week, as I am on vacation.
If some large log has still not been sent, please try a) FTP it to anonymous ftp at ftp.opclabs.com/upload, b) email it to sales09 (at) opclabs.com, c) sent using Dropbox or a similar service.

Best regards