impC001 disconnected; cannot reconnect


#1

Today my impC001, which has been running for several months, disconnected:
2018-11-21 16:32:24 -05:00 [Status] Device disconnected
and is unable to reconnect. When I cycle the power, the dev board LED flashes red, one long followed by two shorts.

I ran blinkup again, and it succeeded (flashed green for 3 seconds), but then went back to the red pattern. According to the information at
https://developer.electricimp.com/hardware/working-with-the-impc001#codes
a failure to connect to the cell network should be indicated by one long red flash followed by three shorts, not the two that I observe. Does anybody know
a) what this pattern signifies
b) if there are any known issues with cellular connections in northeastern N. America?

thanks…


#2

Follow-up: for reasons unknown, the device has now connected again…


#4

Thanks for reporting this. If you provide us with your imp’s ID or MAC address we can take a look at its logs to find out more about its connection history.


#5

Thanks for looking into this. The MAC address is
0c:2a:69:f0:03:9c


#6

We have an open ticket with our provider regarding this; we saw an internal test device be affected in the same way (though it came back on its own), though other test devices were not affected.


#7

Does anybody know […] what this pattern signifies?

Please note we’ll be updating our documentation on impC001 LED behaviour shortly.

On impC001, long red followed by two short reds indicates the imp is attempting to register to a cellular network. On WiFi imps, it indicates it is searching for a WiFi network.


#8

b) if there are any known issues with cellular connections in northeastern N. America?

Yes - our cellular service provider reported some issues over the same period that you were having connection problems. These are now resolved, but we’ve raised a ticket with them to follow up and are seeking better visibility in the event that this reoccurs.


#9

The service provider has now confirmed they had a problem, and are working on both (a) improved monitoring, so they will detect a re-occurrence (b) automated recovery of the affected service component, to prevent device disconnections.


#10

Here is a followup on what seems to be a recurring pattern; there can be periods when the connection is dropped frequently, but it is usually reestablished in less than a minute, and often in just a few seconds:

2019-01-07 10:02:33 -05:00 [Status] Device disconnected
2019-01-07 10:02:38 -05:00 [Agent] device has connected
2019-01-07 12:49:25 -05:00 [Status] Device disconnected
2019-01-07 12:49:54 -05:00 [Agent] device has connected
2019-01-07 13:21:38 -05:00 [Status] Device disconnected
2019-01-07 13:21:39 -05:00 [Agent] device has connected
2019-01-07 14:08:34 -05:00 [Status] Device disconnected
2019-01-07 14:08:35 -05:00 [Agent] device has connected
2019-01-07 20:22:25 -05:00 [Status] Device disconnected
2019-01-07 20:22:25 -05:00 [Agent] device has connected
2019-01-07 22:05:18 -05:00 [Status] Device disconnected
2019-01-07 22:05:23 -05:00 [Agent] device has connected
2019-01-07 22:20:45 -05:00 [Status] Device disconnected
2019-01-07 22:20:55 -05:00 [Agent] device has connected

Sometimes this behavior will persist for several days, then “go away” for weeks.

For my application this is not a problem because the recovery is so quick, but it might pose issues for others.

The long term outage that I reported in November has not been seen since; thank you again for your actions on that.


#11

Can you provide a device ID? We have diagnostic tools that can tell us why the disconnect happened, which will likely narrow this down a lot.

\Note that if you are in RETURN_ON_ERROR mode and you send so much that the buffers fill and you are using a short data send timeout, the device will disconnect - we sometimes see this when people don’t understand those parameters and push too much data at once with a short timeout.


#12

ID was provided earlier in this thread.
I am not aware of any IO taking place when the dropouts occur.


#13

Thanks for reporting these disconnections. After reviewing the diagnostic logs, the imp team recommends updating your imp to the latest impOS 39 version, as we’ve carried out a lot of improvements recently.

Can we go ahead and queue your device for upgrade?

Note that if I do so it won’t actually upgrade until it next reconnects to our servers.

(internal ref. #163144363)


#14

Yes please do. I had noticed that the version it is running is 39.14, dated 31 August, and was wondering if there was a newer release.

Thanks again.


#15

OK, this is done. You should see the imp upgrade when it reconnects, as mentioned above.


#16

Hi Robert, how has the imp’s performance been since the upgrade?

I put a couple of questions about your application into a direct message; if you have time to reply it would help us rule out a number of things that might have contributed to the connectivity issues you were seeing. Thank you!


#17

Hello Harry,

I did respond to your direct message by replying directly to the address from which I received it; evidently that did not work. Please let me know where to send it and I will try again.

thanks,
Robert


#18

You can send to harry@electricimp.com - probably better than the forum messaging system for a conversation!


#19

Sorry about that - I’ve now received your detailed reply (thank you) and forwarded it to the development team.