Hello.
Thank you for the additional information. We were able to reproduce the issue on a small program.

We have made a change that should fix it. Please download and rebuild with QuickOPC 5.72.522 or later (currently on nuget.org and on our Web site), and retest.
The fix was quite delicate. Obviously we have run our normal tests, but there is still some danger of regression. I would welcome your feedback.

Best regards

Thank you. I think I now have an understanding of the issue and your approaches.

First, just some clarifications:

- The fact the WriteMultipleXXXX gets slowed down by *any* of its elements is by design. With the model chosen for returning the results, we need to return from the method only when all results are ready, and that means that in some circumstances, the "fast" parts suffer from the "slow" parts. So if your usage cannot tolerate that, it is correct to search for workarounds or different approaches.

- Even if you do not "address" the disconnected server in WriteMultipleXXXX calls, you have them in your subscriptions, and I suppose it is the reconnections made for the subscriptions that are causing the longer times spent in WriteMultipleXXXX calls. If all you wanted was to reduce the frequency the problem happens, you could do so by setting some parameters we have that control the delay between the reconnections (I can give more info on that if you like) . But you probably want to eliminate the issue altogether.

Now to the actual issue. It seems that there is no "hard" reason why the reconnection attempts made for subscriptions on unrelated servers should cause a delay in WriteMultipleXXXX calls. But there may be an internal implementation reason for it. What we need to do is first to try to reproduce the problem on our side (with some simplified scenario), and then, if reproducible, check if it can be resolved.

So I will work on creating a repro. Please allow some time for it. I will report here when I have an update on this.

Regards

Hello.

I need some more info for clarification :

1. You wrote "...cause write to other running server slowly". Do you mean that the WriteMultipleXXXX method call takes longer than normally, or is there anything else that gets "slow"?

2. Is it my correct understanding that in your WriteMultipleXXXX method call, you combine the nodes from different servers?

3. Can you please elaborate on what have you done when you wrote "Tried isolated client methods". I am asking this because (although I know you do not prefer it) it might be our recommended approach to this.

Best regards