Offline/Online Events

Occasionally, I see a Boron device post an offline followed by an online event. Is there any way to get more information on what the root cause is for these events? Also, is there any way I can collect/copy all the events being posted by all devices?

After looking into this further, I've noticed a consistent pattern where the events occur precisely every 23 minutes. This observation has raised suspicions that I might be experiencing the same issue detailed in this post: Boron LTE Disconnects. It's worth noting that the device operated smoothly for several weeks. Interestingly, when I deliberately initiated additional traffic after observing the offline/online events, it appeared to resolve the issue. I'm curious if anyone else encountered a similar issue and how it can be mitigated on a large scale, especially for devices that mostly remain idle, waiting for functions to be called and this issue appears to be random and sometimes fixes itself after some time. Not sure if load on the tower has something to do with this as well.

A offline followed by online usually indicates the device went offline without telling the cloud about it (non graceful sleep for example). When the device talks to the cloud again, the first session is closed, the device marked as offline, and immediately online again.

That’s pretty normal. You can grab the event stream to monitor this.

The 23 minute period aligns with the device’s keepalive. Every 23 minutes a device sends up a message so the cloud knows it’s online.

In your case, it seems the device falls offline prior to this message being sent. Are you able to communicate with it before the 23 minute keepalive?

Please DM your deviceID to @Support so that the team can take a look.

The device was working fine for weeks and out of nowhere experienced this issue. I was not able to communicate with it before the 23 minute keepalive. After seeing the offline/online messages I immediately called a couple exposed particle variables just increase the data connections and that seemed to have fixed the issue. I have seen this happened in the past and rickkas7 explanation made sense.

1 Like