Professional OPC
Development Tools

logos

Subscription stopped firing events

More
23 Jul 2019 19:49 #7559 by sjscheider
Great. I have been testing the new version for the last 2 days without issues.

Thanks for you efforts on this issue.
The following user(s) said Thank You: support

Please Log in or Create an account to join the conversation.

More
21 Jul 2019 09:22 #7549 by support
Hello.

I was able to reproduce the issue, and fix it. In fact, it *was* the issue I suspected, the one fixed in 5.54.1151.1. However, due to a mistake in the build procedure, the fix remained in .NET Framework code and was not automatically merged into .NET Core part. This explains why 5.54.1151.1 resolved the customers' issue (they were using .NET Framework), but you were getting the very same problem even with later builds - that's because you are using .NET Core.

Please download and rebuild with 5.54.1296.1. This is now on our Web site and on NuGet.

Thank you for reporting the issue and providing valuable information.

Best regards

Please Log in or Create an account to join the conversation.

More
18 Jul 2019 15:01 #7547 by support
Very good. Thank you. I will continue as planned, but this information brings me much closer.
I will let you know here when I have an update.

Please Log in or Create an account to join the conversation.

More
18 Jul 2019 11:46 #7546 by sjscheider
It worked after needing idle for over 6 hours.

Please Log in or Create an account to join the conversation.

More
18 Jul 2019 04:03 #7540 by sjscheider
I changed QueueCallbacks to false and this appears to fix the issue.
_easyUaClient = new EasyUAClient { Isolated = false, QueueCallbacks = false };
I'm guess we do not want to run with QueueCallbacks set to false.

I've only run our app for about 30 minutes, but before I was able to reprocess the issue after only about 10 minutes of inactivity. I'm going to leave it run overnight to see how it performs.

Please Log in or Create an account to join the conversation.

More
17 Jul 2019 15:02 #7539 by support
Thank you very much. This sounds very good. Here is what I plan to do:

1. Try to reproduce the issue without your tools first. The reason why I haven't done that yet is because (a) I was assembling the necessary information, and also (b) because we have done similar tests and they appeared to work. It needs to be seen whether e.g. NET Core plays a role (most of our testing is done under .NET Framework).

2. If I fail to achieve a repro myself, I will ask you for your tools and we will figure out a way how to deliver them.

Best regards

Please Log in or Create an account to join the conversation.

More
17 Jul 2019 14:57 #7538 by sjscheider
I'll try to answer all of you questions.
  1. We are running using .NET Core, 2.2 under Windows. We have experimented with it under Linux but for our initial release we will be running under Windows 10.
  2. No part of the system is under heavy load. We are able to duplicate this issue when we are connected to a single device and subscribing to a single node.
  3. We are able to duplicate the issue all the time. If data change on the device, we have no issue. If data does not change on the device the issue occurs.
  4. We have a standalone .EXE that you can run (no install needed) that simulates the device the we are connecting to. This is how we are able to run against 100s of devices. We actually have on server PC running 100 copies of the simulator without issues. (I think I customer has even more running on a single PC.) This simulator can run using a few different scripts which mimic how the device works, causing data changes on the device which result in data change events being sent to the client. We are able to duplicate this issue by running no script which results in no data change events occurring. This mimic what we are seeing in the field against a few physical devices that are basically idle. I can send you this app and instructions on how to use if this is something you think is worthwhile.
  5. We also have a simple .NET Core console app that will communicate to a single device using a single subscription. We can share this as well. We just need a way to share this outside of a public forum.

I will run setting the QueueCallbacks to false and retest later today and let you know what the results are.

Please Log in or Create an account to join the conversation.

More
17 Jul 2019 12:22 #7536 by support
One more question please:

Somehow I was under impression that it is known that you are running under .NET Core, but I cannot find an evidence for it in this forum thread. So, are you running your app under .NET Framework, or .NET Core? If it is .NET Core, is it on Windows or Linux?

Thank you

Please Log in or Create an account to join the conversation.

More
17 Jul 2019 08:51 #7533 by support
The reason I was concerned about the build numbers is as follows: We had an issue with similar symptoms: Things were working OK but notifications stopped if there were no data changes. The bug affected, however, only the notifications being sent from QuickOPC to users code. Everything worked fine in between QuickOPC and the server. We fixed the bug in build 5.54.1151.1. The customers that had the issue then reported the fix had resolved the problem.

So, it is really critical to use the newer builds. But according to your report, you have them already.

The similarity with this issue is however something that bothers me. It probably is *not* the same thing, because there are important dissimilarities as well (e.g. the fact that it behaves differently in the debugger). In order to test this, can you please set the QueueCallbacks property on your EasyUAClient object(s) to 'false', and re-test? Plan: if it helps, we are dealing with a variant of the old issue. If it does not help, it is something truly different.

--

The fact that there are event entries about publishing stopped/recovered is suspicious, although by itself it should not cause the particular issue (it is a warning; when it escalates to a problem, your code should be notified - so from my point of view, the first issue here would be the absence of the notification, and not the actual "publishing halted" issue). Is any part of the system under heavy load (the server, the network, the client side - CPU etc.?)

--

You described that/how you were able to reproduce the issue in a constrained environment. When you do the same repeatedly, does it *always* happen, or just sometimes?

--

Can you please elaborate on what you mean by "sharing our simulator with you", and the test app? Are you saying you have two pieces of software that we can run here and try to reproduce the issue?

Best regards

Please Log in or Create an account to join the conversation.

More
15 Jul 2019 15:11 #7527 by support
Thank you. Both these builds (5.54.1156.1 and 5.54.1281.1) should be OK. I just wanted to be sure that you were not running something older (in which case there would be an explanation for the issue).

Please Log in or Create an account to join the conversation.

Moderators: support
Time to create page: 0.201 seconds

      

 Recommend this on Google