Repository navigation
Replies: 1 comment
|
Update after three more dropouts: the one-way failure is now measured automatically, on a second lamp too Since the first post I have ruled out two things and added a test. Ruled out
New test: a command sent during the dropout An automation now waits until an Innr lamp has been unavailable for 10 minutes and then writes a harmless setting to it ( First result, lamp B (a second RCL 232 C in another room), 16 November:
So the write sent in the middle of the dropout was received and stored. The lamp was on the network and listening for the whole 94 minutes; only its own messages did not reach the coordinator. This matches the manual on/off test on lamp A in the first post. Recovery pattern All three self-recoveries I have timestamps for happen at the lamp's own periodic OTA query (every 5 minutes, at a fixed offset per lamp): lamp A at 10:27:06 and 20:52:06, lamp B at 04:33:03. The offset is unchanged before and after, so the lamp did not reboot. One of its own periodic messages simply got through again. Frequency, past 7 days
Question Does this look like the lamp holding a stale route to the coordinator and not repairing it until something forces a new route discovery? Is there anything I can capture during the next dropout that would confirm it, for example a sniffer trace, or a specific Z2M debug log? |
Uh oh!
There was an error while loading. Please reload this page.
Hi all,
Three Innr ceiling lights (RCL 232 C ×2, RCL 231 T) in my network regularly go "offline" in Z2M for 10 minutes up to 13 hours, while the other ~60 routers are stable. Innr support has been very helpful and suggested I ask here; they'll follow this thread.
Summary of what I found
on, waited 10 s, sentoff. The lamp switched on and off, but Z2M got no response to either (Timeout after 10000ms).queryNextImageRequeststops arriving.So it looks like the lamp's route back to the coordinator (many-to-one / route record) goes stale, while everything else keeps working.
Questions
Setup
Dropout statistics
Over 15 days (late Oct – early Nov): ~35 dropouts for lamp B, 13 for lamp A, 5 for lamp C. Most happen with no commands or remote use in the 15 minutes before, also at night. Power cycling fixes it immediately; otherwise the lamp recovers by itself after 10 minutes to 13 hours, without rebooting. After I stopped the 30 s polling, dropouts returned within 2–3 days.
Network maps during dropouts (lamp A)
Note: the Innr RCL lamps never answer LQI / routing table requests, also when they work normally, so I can't see their own tables.
Raw network maps (JSON) available on request.
Thanks for any pointers!
All reactions