[DEFECT] Admin Panel Unreachable on Some Nodes Despite Being Visible in Dashboard

[tyronechrisharris]

Issue

Description

The OpenSensorHub admin panel is not accessible on some build nodes, even though those nodes are visible and reporting status correctly in the dashboard. This creates an inconsistency where the system appears healthy at a high level, but administrative access via the web UI fails on specific nodes.

Two affected systems exhibit this behavior consistently. A screen recording demonstrating the issue is available here:
https://youtu.be/ESbRxQ36L5o

Log files from both affected systems have been attached and show recurring Jetty websocket proxy errors and MQTT TCP socket read failures, which may be preventing the admin panel from loading or maintaining a connection.

---

Steps to Reproduce

1. Deploy multiple nodes with multiple lanes for one week.
2. Open the main OpenSensorHub dashboard.
3. Confirm that multiple nodes are visible and reporting status normally.
4. Attempt to access the admin panel (web UI) for each node.
5. Observe that some nodes fail to load or are unreachable via the admin panel, while others function as expected.

---

Expected Behavior

All nodes that appear online and visible in the dashboard should have a reachable and fully functional admin panel.

---

Actual Behavior

On certain nodes:

• The admin panel does not load or is unreachable.
• The node still appears healthy and visible in the dashboard.
• The issue is reproducible and node-specific.

---

Screenshots

Screen recording demonstrating the issue:
https://youtu.be/ESbRxQ36L5o

---

OpenSensorHub Config File

Not attached yet. Can be provided for both affected and unaffected nodes if configuration comparison is helpful.

---

OpenSensorHub Data Store File(s)

Attached log archives from two affected systems:

• moduledata.zip (affected node A)
• moduledata (1).zip (affected node B)

moduledata (1).zip
moduledata.zip

---

Additional Information

Observed log patterns common to both affected nodes:

1. MQTT TCP socket read failures

• Repeated errors such as:

  • ERROR - Error reading data from MQTT TCP socket
  • java.io.IOException: An established connection was aborted by the software in your host machine

• These suggest the TCP connection is being interrupted at the host or network level (e.g., firewall, proxy, NAT, or connection instability).

2. Jetty websocket / proxy errors

• Repeated errors including:

  • ERROR - Internal websocket proxy error
  • org.eclipse.jetty.io.EofException
  • Caused by: java.io.IOException: Connection reset by peer

• These errors appear to coincide with attempts to access the admin panel.

3. HTTP server startup appears normal

• Logs indicate the web server starts successfully, e.g.:

  • HTTP server started on port 8282

• This suggests the admin service is launching, but websocket or TCP-level failures may prevent the UI from functioning correctly.

Notes / hypotheses:

• The issue appears node-specific, not global.

• Nodes with inaccessible admin panels still report status correctly to the dashboard.

• Possible contributing factors include:

  • Websocket proxy handling
  • Network/firewall differences between nodes
  • Port exposure or binding issues
  • MQTT connection instability impacting the admin UI