We are using client that sending up to 20 short packages during first 2 minutes on top hour (top hour means hour:00 ).
We used https://v2console.thethingsnetwork.org/ long time and we never seen any problem . All packages arrived to application device page and then to our server without any lost.
Now we decided to use chirpstack server. Installed, configured - packages started to arrive finally on our server.
But many packages lost - about 7-10 packages from 20.
All packages arrived to gateway-bridge service. I can see all of it on that Gateway page.
But then many of those packages not arriving to application-service.
Next top hour situation repeated. Just first 7-10 packages arriving on destination. Seems that problem in some time delay on network-service.
– we using 9 different frequencies and on network-service configured 9 channels for each frequency. And I can see that any package with first time (after top hour) visible frequency passing. But if then arriving in some 6-20 seconds other package with the same frequency - it lost.
Looks as that on network-service channels still busy long time. I cannot see any parameters on configurations of services about how long those channels are busy and how fast it emptied.
Can somebody help for solve problem ?
I’ve tried to increase quantity channels on network with repeated frequencies. But no any effect. Tried to play with some other parameters ( really not found any that can help for busy time of network-service channels) but I cannot see any effect.
Thanks in advance for any help or ideas where necessary to look possible solution…
Did you look into the (error) logs of the Gateway Bridge / Network Server and Application Server?
Yes, I’ve seen messages in chirpstack-network-server journal that for arrived package not found channel. About the same messages were when not all possible frequencies were configured as channel. But as we used 9 frequencies only 6 additional channels.
But week ago I’ve added chirpstack-multiplexer with option “uplink_only=true” and now I can see situation much better - still not all messages arrived, but not between “gateway bridge” and “network-server”.
But still that question about those channels in network – how long it’s busy ? How many events per minute can be managed by those gateway-bridge and network services ? If we want to use thousand clients (distributed in many km2 fields with 20 GWs) sent messages during 20 minutes – can GW bridge and network manages all packages ?
There’s no such concept.
Of course at a given gateway, a given frequency and spreading factor can only carry one message at a time, so one must finish before another can begin. If two nodes try to transmit at the same time with the same air settings, they’ll unavoidably interfere with each other at any gateway where they overlap without drastically different signal levels.
But there’s no “collision” in the infrastructure between messages that are actually unique and successfully received by at least one gateway in the system.
Are these valid LoRaWAN uplinks each having a unique uplink frame count higher than the previous?
(The only “busy” accounting in software is that nodes are required to not re-use an uplink channel too frequently, and in some jurisdictions gateways are also required not to re-use a downlink channel too soon; but neither of those would result in the behavior you are reporting since the first mechanism would be about the node never transmitting rather than transmission being ignored, and the second would be about downlinks unsent but have no impact on the preceding uplink they could have been sent in response to if not prohibited)