Before answering, I’m making some assumptions here …
- This screenshot comes from a dashboard time series graph, and each of your segments targets a single device/attribute combination.
- When you say “device disconnect node”, I’m guessing you actually mean a Device: Inactive Trigger, given your followup statement of “goes off after a couple minutes without a mqtt message packet being received.” (A Device: Disconnect Trigger, by comparison, would fire immediately as soon as a device disconnects from the broker.)
If my assumptions are accurate, then the gaps you see in the graph could be reconciled against the lack of an inactivity alert if the the attribute for a given device was not reported for that gap time period, but other attribute values were reported for the device in that time period. When returning graph data to the dashboard, if we do not have data for a given resolution bucket, we simply return no data for that resolution bucket’s timestamp. (If, in the graph, you’d like these gaps to connect through anyways, you can set that option per segment.)
Alternatively, depending on how long “a couple minutes” actually is (2? 5? 10?), it could be possible that your device stopped receiving state reports for those gap periods, but the gap is not long enough to trigger the Inactivity Trigger.