I’m in the middle of setting up a bunch and noticed huge variation in the listed delay from sub-minute to 30+ minutes even when the sensor is still showing up on the raspberry pi as a w1 bus device. In at least one case, the raw value shown in the home assistant page didn’t match the raw value in the w1_slave file for quite a long period before eventually updating again and getting back in sync.
Is it maybe that if the sensor doesn’t observe a change in temperature it doesn’t generate an event thus the timestamp of the last temperature read doesn’t change? Although holding the sensor to heat it up doesn’t seem to reliably trigger home assistant to update it’s value in the UX. And I’ve seen a case where a sensor was still updating it’s value but the file timestamp on the w1_slave file (and others) was still from the previous day. While at other times, the timestamps update when the reading gets updated. So if there is some kind of eventing going on, it’s not reflected in the file timestamps for the devices in the linux filesystem.