Hi,

Some servers sometimes do not have their nodes available immediately after reconnecting. I would say that is probably a bug and it should be reported to the server vendor. Though, depending on the server it might also be somewhat valid behavior depending on the data source (e.g. if it aggregates data from multiple sources, it might not have yet had a connection to all of them, and this might not be deterministic, so it might just have to accept incoming connections after some period).

That version of the SDK is quite old, I would recommend updating it. https://downloads.prosysopc.com/opcua/Prosys_OPC_UA_SDK_for_Java_4_Release_Notes.html, 4.6.0 added: “Improved: Subscription.setRecreateFailedItems(boolean) can be called to enable a Subscription to automatically attempt to recreate MonitoredItems with error code Bad_NodeIdUnknown since this StatusCode can imply in some cases that the Server hasn’t added that Node to its AddressSpace yet.” and in 4.8.0 that is by default true, due to these cases. And 4.9.0 is the latest version, thus I would recommend updating to that.

I should note that it could also be something else, but the above was the most typical case. For example, if you have made your MonitoredItems via NodeId, the index could have been changed on the server (per spec it can change every Session, only the 0 and 1 indexes are fixed, but like typically servers keep it static). This could be solved by making the item via ExpandedNodeId instead (by using the NamespaceUri version).

And also in theory the node could have been removed, there is no way for the SDK to know that, but I assume this is not the case here as you can validate that it probably exists.