Does the Tempest have memory?

I thought I had understood that the Tempest has a bit of memory so that when it looses connection with the hub, it can still collect data and sent it when it comes online again. Is this true? This morning I woke up to two mails (and two notifications on the phone) that the Tempest was offline and that it came online again. Not seen that before, well, maybe once but it’s not a regular thing.

Thankfully I had recently updated my weewx driver to register RSSI values. The graph looks like this:


(Curious that the wifi signal seems to get much more stable while the station is unreachable)

The mail that it got offline is from 1:30. The station seems to get offline 0:25-ish (5 minute resolution in my weewx-db and the timestamp is from the last reading, so that must have been 0:20-0:25) and that corresponds to the last reading I find in the Tempest app, which is 0:21. I suppose the mail comes an hour-ish after it goes offline.

The mail that it got back online is from 7:40 and that corresponds with my weewx-data which puts it between 7:40 and 7:45. The Tempest app has data from 7:27.

As you can see on the graph, there are also a few problems just before the main blackout, from 0:10-0:25. The Tempest app does not have this gap however, so I suppose that was filled back.

But does this mean this memory is only 15 minutes? That’s rather a lot shorter than I’d expect, to be honest :slight_smile:

BTW the battery was fully loaded earlier that day so I’m pretty sure it didn’t go offline because of an empty battery.

And as you can see I placed my hub in a bit better position a bit after 21:00 :slight_smile:

it has memory to keep several hours of data. The hub has memory as well (in case the internet connection drops out)

Hmmm, why doesn’t it seem to show up in the data I have then? Shouldn’t that data be available somewhere by now?

That’s a different question. Why didn’t it keep on measuring and storing data?
RSSI of -65 shouldn’t be a problem. -85 is borderline.

Yeah, it obviously had a bad connection, which is why I moved my hub. But if what you are saying is correct and it has this memory, shouldn’t some of that data have been recovered somewhere?

I thought the distance between hub and sensor could be 100m or so, but I couldn’t find that number after a short search, and even if I could, I suppose it isn’t valid for a well isolated Norwegian roof of an apartment building that is not that old (2016).

But as I said, until now there haven’t only tiny dropouts, this was the first time I saw such a long period without signal.

for sure it has the memory and I would expect it to fill some of the data, but it clearly didn’t. Perhaps it stopped measuring all together for some reason, I don’t know. It seems to be working now. Keep an eye on it.

Hi @jjvdgeer ,

I do not see a link to your station on your profile so I can not look at your data.

My Sky does backfill following a disconnection but it does not fill my weewx database automatically.
I have a web site displaying my weather data in my own graph style because it works faster than weatherflow graphs and displays in my preferred style. Sometimes they might fail for some reason so I include a link from those graphs back to the weatherflow graphs because when my graphs or my weewx fails the weatherflow graphs are still working AND they back fill automatically.
cheers Ian :slight_smile:

Also it is possible to have a hub that fails on certain wifi channels. I have experimented changing wifi channels due to it once causing my problem of loosing connection. Following my experiments and analysing nearby wifi activity and considering possible 2.4GHz wireless cameras nearby I now have set my wifi to a fixed channel that works well for me. If I leave it on auto it sometimes looses connection when it selects a bad channel.
cheers Ian :slight_smile:

Not 100% sure but I think it only backfills to the servers. The local data stream via UDP doesn’t do it but I think I need to confirm this with Rob.

That’s why I verified with the Tempest app. I suppose the app creates its graphs with data from the server.

BTW @iladyman I added my station to my profile :slight_smile:

1 Like

I remember having a power failure for a couple of hours which stopped my hub and my wifi from working. When the power came back on some of the data for some of the sensors (Sky,Air,Tempest) backfilled ok over a period of time. Some of the data was lost.

Had a return, when the tempest backfills it does so on all streams.
On UDP it is known there can be some duplicate records.
The question is how does weewx handle events coming in with data stamps in the past ??

That’s a good point, I don’t know. It may actually work well and maybe that’s exactly what happened for all I know. Hm. Difficult to know I guess… Although… I suppose (but that’s a guess) it doesn’t backfill the device_status messages, so if it did that I should have records without RSSI data but with data for the other sensors. That doesn’t seem to have happened.

Also, the main developer of weewx seemed rather negative about the Tempest because of the fact that it could in some situations sent out-of-order packages. I suppose if weewx handled that fine, he wouldn’t be so upset about that, but that’s again a guess. (see Backdated UDP `obs_st` messages - #24 by rderr and )

It doesn’t tend to care much either way.

What it doesn’t like is duplicate timestamps (secs since the epoch), and it’ll almost certainly discard anything with a timestamp it’s already got in the database. This is logged in the syslog pretty clearly, for what that’s worth, but it’ll discard them regardless.