Somtimes my Aeon ZW100 shows very low temperatures. (Below 100 degrees celsius) On some ZWave sensors I also get temperatures that are much too high. Is this a problem of the sensor or of the openhab addon? Here is a screenshot from today:
@chris Any idea what could cause this?
I would suggest to check the logs to see what the device is sending, but this is a known issue with these devices so I’m pretty sure it’s not the binding but the log would confirm.
Hi Chris, this problem only occured two times this month on this temperature sensor, so I would have to run a debug log for a long time. But if this is a known problem with some ZWave seonsor then I guess it’s that.
It’s a known problem with the Aeon sensors at least. In the past it’s often triggered by polling, so turning polling period down might help.
At the moment the polling of the device is set to 30 minutes. What setting would you recommend? And would you recommend a higher setting for all ZWave devices in a large network?
It’s up to you really. However in large networks, I would certainly consider increasing it to something like 2 hours. As a simple calculation, you can do something like (number of devices, multiplied by number of channels) should be less than one per “5 to 10 seconds”. This will keep the background level of communications low enough to ensure there’s no significant impact. Of course, in reality these tend to come in bursts, but at least it’s a reasonable starting point.
So, if you have 100 devices, and each device has 5 channels, then that’s 500 requests, so I would keep the number above 2500 seconds, so ~1 hour minimum.
For problematic devices, just wind it up to 1 or 2 days 
Ok thanks for the detailed explanation. One last question, what exactly is the downside of polling less often? Is this just to check if the device is still online or is there more to it?
Yes - I think that’s really the only reason. All normal traffic should be handled by other means (eg associations for reporting sensor changes, status updates, or alarms) so polling is just there to make sure that if you don’t switch on a light for a week, if it dies, you know about it.
I’ve now set every of my devices to 6 hours polling. This also seams to reduce the cancel messages a lot. Before I had an average of 8 cancel messages per hour. Now the average is only 2 cancel messages per hour.
