Thank you for your response. Hopefully you can help us out with these follow-up questions:

Question 1:
The code you posted:

object.getEventNotifier().contains(EventNotifierClass.SubscribeToEvents);
where object is an UaObject.

Detects, in our case, the Server object and objects under Server. However, it does not detect a custom event created in our server address space. Our custom event type is created in the same way as “MyEventType” in the sample client in the Prosys Java SDK, and is located under Types/EventTypes/BaseEventType/. Is this expected behavior? If so, how can you generically detect custom events, such as “MyEventType”, created by the server?

Question 2:
Is the Server object an address space node required by the OPC-UA standard? What would normally be the interesting information in the Server object?

Question 3:
What is the proper way to detect variables that allows listening to data changes? Right now, we detect such variables the following way

if (referenceDescription.getNodeClass().toString().equals(“Variable”)) {
// setup subscription
}
where referenceDescription is of type ReferenceDescription.

This works, however, is there a better way? In the sample client in the Prosys SDK there is the following code:

UaNode node = client.getAddressSpace().getNode(nodeId);
if (node instanceof UaVariable) {
// Validate that history is readable for the node
…

When using the conditional above, only some variables are detected in our server address space, NOT including custom created variables of type PlainVariables by the server:

final NodeId temperatureId = new NodeId(nodeManager.getNamespaceIndex(), “temperature”);
temperatureVariableSensor1 = new PlainVariable(nodeManager, temperatureId, “temperature”, Locale.ENGLISH);
temperatureVariableSensor1.setDataTypeId(Identifiers.Double);
temperatureVariableSensor1.setCurrentValue(99.);
device.addReference(temperatureVariableSensor1, Identifiers.HasComponent, false);

Is this expected behavior that these variables are not detected by checking for type of UaVariable?

By the way, how do you get code highlighting in the posts?

Thank you.

You are correct, all variables do appear is UaVariable nodes. I have two additional follow-up questions

1. Exactly how do you link the custom event type to the Server object in order to have the event type detected by:

We tried the following without any luck:

Where server is of type UaServer and customEventType is of type UaObjectType. Does Identifiers.Server refer to the server Object?

2. What is the proper way to clean up resources after a connection failure is detected on the client side? Right now, on the client-side, we only call disconnect() on the the UaClient object. Is this sufficient? Should we also explicitly unsubscribe to all subscriptions, or will all resources related to subscriptions be released when the subscriptions time out?

Thank you.

If you choose to keep the connections on the server side with UaClient.setKeepSubscriptions(true), does this mean that when the client connects to the server again there is no need to browse the address room and actively create subscriptions?

Moreover, we have a few questions regarding timestamps:

1. For events, there is a default attribute, “time”, that provides a server-side timestamp. In our OPC-UA server, based on the sample server, this attribute appears to always be set. However, for data changes on data variables, such as PlainVariable, is there an associated server-side timestamp? The DatainnMonitoredDataItemListener provides a onDataChange() callback with a parameter of type DataValue. This parameter has a getServerTimestamp(), which appears to be null all the time. Should we be using this timestamp? Is there a miss configuration on our OPC-UA server since it is always null?

2. In addition to getServerTimestamp(), the DataValue parameter mentioned above also has a getSourceTimestamp() method. What kind of timestamp is this? A timestamp representing when the client SDK received the message from the server?

Thanks, exactly what we needed. We have stumbled upon three more questions:

1. Is it possible for the server (of type UaServer) to bind to a arbitrary available port, and then retrieve that port for the client to connect to? We tried binding to port 0 using server.setPort(), but server.getPort() returns 0 and the server fails. We would like this functionality during integration tests.

2. During browsing of the address space, we encounter several nodes that have the same path in the address space, but different NodeId. For example, we encounter two UaVariables at “Types/ObjectTypes/BaseObjectType/NamespacesType/AddressSpaceFile/GetPosition/InputArguments/OutputArguments”, one with NodeId 11691, and the other with 11692. We assume this node path is default provided by the UaServer. Is this expected behavior by the UaServer? If so, then the node path is not unique by itself?

The code we use for browsing is:

where client is of type UaClient, and nodId of type NodeId.

For detecting data variables that provide data change events, we use:

where node is of type UaNode.

3. Given the path in question 2., what would be the usefulness listening to data variable changes under that path? I guess it will not give us much. Is there any specific address room paths that are especially useful to listen to in the OPC UA standard, including both events and data variable changes? Events under the Server object at root is probably useful?

Thanks. 3 more questions

1. Can TranslateBrowsePathsToNodeIds() be used to directly look up a node in the address space? For example, let’s say I want to directly look up an event at the path “/Objects/MyNode/MyEvent”, how would I do that in code? Or do you have to find that node by traversing from a known root node, such as Identifiers.Server.

2. Could you comment on our current method of browsing the server address space? Our approach is as follows. We have a generic depth-first traversal of the entire address space. Initially, every path in the address space is ignored (not traversed), however, we can configure the traversal using a “whitelist”, which defines which absolute paths (including leaf and sub-trees) should be included in the traversal. If a traversed node in the address space is either a data variable or event node, a subscription will be created for it. This way it is easy to include data variable and event nodes that are know before the traversal, as well include sub-trees that might have interesting nodes that we would like to create subscriptions for. Does our approach sound like a complete overkill, or reasonable? Would it be more sane to look up individual nodes using absolute paths, instead of a generic traversal?

3. Related to the above questions. We have experienced significant delay when traversing nodes in the address space when communication is over GSM. Our guess is that the network latency is the primary cause. Are we correct? What kind of actions would you suggest in such a case? Perhaps using TranslateBrowsePathsToNodeIds() to directly access nodes, if that is possible, and hence avoiding traversal latency?

Thank you.