Copyright ©

Mongoose OS Forum

ATTENTION! This forum has moved to:

Do not post any new messages.

Determining if os/firmware crashes remotely.

siftd106siftd106 United States

Is there currently a way to determine if the os/firmware crashes and reboots.
I would like to keep track of how often this happens for our fleet of devices, but do not currently know of a way to determine if the boot was a fresh power up, or a reboot due to crash, or a planned reboot.
I am imagining something to check during the boot, that I can then queue up a message to be sent to our mqtt server so I can report on this. It will help us identify when we implement new features, and the new features have some level of bug in them.

Thanks in advance!


  • ulsoulso Stockholm
    edited March 5

    Wouldn't this be a perfect use for the MQTT "will" message?
    I haven't tried using it my self yet, but maybe you could try this

    mos config-set mqtt.will_message="Connection lost"

    and see if you get that MQTT message when the device reboots.

    I may be wrong but I think the the "will" message setting tells the MQTT server what message you want to be sent when the connection to the device is lost.

  • siftd106siftd106 United States

    Very interesting, I like the out of the box thinking on this one.
    We are using AWS IoT for out MQTT implementation, and I don't think they offer a "will" message.
    They may offer some other form or getting the same info, like some type of connection info that we query on a consistent basis from the backend to see when it disconnects, or stops "ping"ing.
    I'm not sure that knowing when a device disconnects will get me the info I need, since a disconnect doesn't mean a reboot/crash. Our devices are often installed in locations that have very poor wifi signals, and connections are lost, re-acquired often.
    I also am currently getting a message everytime the devices acquire an ip address, and send that info along via mqtt. This lets me track how often the network re-connects, which is another very useful bit of information.

    I wonder if I can hook into the system time to achieve this info. Does the system type get reset to epoch every time a crash happens. I know that it will maintain the system time somewhat accurately on a planned reboot, but I am not sure about after a crash.

    Does anyone know if there is anything different between a reboot, crash, and a fresh boot that I may be able to look at to determine if when it is a fresh boot vs a crash?


  • ulsoulso Stockholm

    I know that for instance the CC3220 SDK has a function called PRCMSysResetCauseGet() that you can use to find out the reset reason. The ESP32 seems to have something similar in the rtc_get_reset_reason() function that is discussed in the following thread

    If you happen to use a CC3220 or an ESP32 you might try to call one of those functions as soon as your fw is starting, to see if there's maybe reason to take special action.

  • siftd106siftd106 United States

    Ohhh nice find. We are using the ESP32. I'll check this out to see how the reset reason differs between crashes, powerups, and system reboots.


Sign In or Register to comment.