A Node-RED OPC UA event subscription to a SINUMERIK server can deliver the initial alarm burst and then leave the server unreachable. In the reported case, changing the ActivateSessionRequest LocaleIds from an empty array to ["en-us"] did not prevent the failure, while the same event access worked with UaExpert. Siemens was contacted and identified the remedy as a correction in a newer OPC UA server version; the exact fixed version was not provided.
Separate the Trigger from the Root Cause
Starting the Node-RED event flow triggers the observed failure, but the server crash is the defective behavior. A conforming diagnostic approach must distinguish the client request that exposes the defect from the server-side defect itself. The successful UaExpert test shows that the server can provide these events through another client path, but it does not prove that both clients generate identical subscriptions or requests.
The evidence does not establish a specific malformed request, event-count limit, or LocaleIds defect. About 30 messages arrive initially, but no supported threshold connects that count to the crash. Treat request formatting and initial event volume as investigation inputs, not confirmed causes.
Record the Failing Event Configuration
| Item | Reported value or behavior | Engineering conclusion |
|---|---|---|
| Server | SINUMERIK 840dsl OPC UA server | Primary reported affected system |
| Event root | ns=2;s=Sinumerik |
Preserve for reproduction |
| Event type | i=2041 |
Preserve for request comparison |
| Client action | events |
Failure occurs on the event-subscription path |
| Configured time | Recorded flow setting; its role is unproven | |
| Security | None |
Reported test configuration |
| Initial delivery | About 30 messages | Delivery succeeds before the server becomes unreachable |
| LocaleIds test | ["en-us"] |
Did not resolve the failure |
| Alternate client | UaExpert works | Use it as the comparison baseline |
Later reports mention similar behavior with Siemens 840SL using OPC-UA v2 and Siemens ONE using OPC-UA v3. These reports broaden the systems worth checking, but they do not establish that every installation or build is affected.
Diagnose and Escalate the Crash
- Reproduce the failure with the recorded event root, event type, action, timing, and security settings. Capture the client console output and the point at which the server becomes unreachable.
- Run the equivalent alarm and event access with UaExpert. Record that result as a working baseline and compare the subscription behavior with the Node-RED path.
- Do not treat
["en-us"]as the fix; that change was tested without success. Do not infer an event limit from the initial burst unless server diagnostics identify one. - Provide Siemens support with the failing flow configuration, console output, server identification, installed OPC UA server version, and the successful UaExpert comparison. Request the exact corrected build for the installed control.
Direct PLC communication is not an equivalent workaround when the requirement is to retrieve NC alarms and events. The evidence explicitly distinguishes those NC events from PLC data.
Apply and Verify the Server Correction
Siemens reportedly planned to correct the problem in a newer OPC UA server version, but the evidence supplies no release identifier or compatibility details. Confirm the applicable corrected version through Siemens support before updating the control.
After installing the approved server version, restart the same Node-RED flow and verify that the initial messages arrive, the event subscription remains active, and the OPC UA server stays reachable. Repeat the UaExpert test as a control. If the crash remains, return the captured logs and exact installed version to Siemens; no evidence-supported client-side workaround is available.
FAQ
Why does the SINUMERIK OPC UA server crash after Node-RED starts events?
The flow triggers the failure, but the evidence points to a server defect that Siemens planned to correct in a newer OPC UA server version. No specific request field or event-count limit was confirmed as the root cause.
Does setting LocaleIds to en-us fix the SINUMERIK event crash?
No. Sending ["en-us"] instead of an empty LocaleIds array was tested and did not stop the reported crash.
What should I send Siemens support for this OPC UA failure?
Send the server and installed OPC UA version, the flow values ns=2;s=Sinumerik and i=2041, client console output, the approximate 30-message startup burst, and the result showing that UaExpert works.