As the update to v94 was missed for the first night of several stations, maybe someone from weatherflow can give us an insight view of the update process ?
see WF response later on in the thread New Hub Firmware: v94 The updates are staggered over a few days.
4:40p Pacific / 11:40 UTC : v94 - 90.9% ; v91 - 8.6% ; still rolling out.
2 Likes
The question moreover pointed onto the technical background…
Are the hubs polling is the server pushing …??
It is most likely a combination of both.
Since the Hub does not have a public IP address and is in constant contact with WeatherFlow servers, I think there is an exchange that allows the server to push an update to the Hub.
Sigh - I missed my window here as mine’s updated to v94, but it would be pretty interesting to try to capture that traffic with Wireshark. Presumably you’d look for:
- traffic originating from the hub
- that isn’t to broadcast udp/50222
- that isn’t a simple arp for the network gateway
I have to think the Hubs must phone home ‘occasionally’ with some kind of traffic to look for software and/or tuning updates…
The Hub is in constant contact with the servers. As often as every three seconds.
DNS queries in the last week or so from the hub according to my pihole logs:
NTP-related queries:
* 0.weatherflow.pool.ntp.org (and 1. and 2. and 3. hosts as well)
* time.nist.gov
* time.weatherflow.com
Miscellaneous queries:
* mqtt.weatherflow.com
* s3.amazonaws.com - guessing the firmware update might be from there ultimately
The most frequent DNS lookups by far are to mqtt.weatherflow.com so I’m guessing that’s the host (or gateway) the hubs actually talk to, especially given the 300 sec TTL on the DNS settings.
Not that it matters really for more than just the adventure of figuring stuff out…
It would be kinda cool to decipher the software update transactions, for me at least coming from an avionics background at former $work where I had to get into dataloading for airplane LRUs a bit. Interesting stuff.
3 Likes