We use cookies to try and give you a better experience in Freshdesk.
You can learn more about what kind of cookies we use, why, and how from our Privacy policy. If you hate cookies, or are just on a diet, you can disable them altogether too. Just note that the Freshdesk service is pretty big on some cookies (we love the choco-chip ones), and some portions of Freshdesk may not work properly if you disable cookies.
We’ll also assume you agree to the way we use cookies and are ok with it as described in our Privacy policy, unless you choose to disable them altogether through your browser.
Hi Milesight team,
We're running a UG65 gateway with the built-in network server (/usr/bin/loraserver -c /etc/loraserver/loraserver.json, log reports starting LoRa Server ... version=8.0.1).
We're building a time-critical Class-B multicast application (an uplink alarm triggers an LED actuation pattern on several end nodes), and we've traced a fixed ~2.05 s delay between the moment our application enqueues a multicast downlink via the REST API and the point the network server actually requests the first Class-B ping-slot — on top of the normal 0–960 ms wait for the next ping-slot itself.
We dug into this as far as we could from our side:
- /etc/loraserver/loraserver.json mirrors the field names of the upstream open-source brocaar/chirpstack-network-server Config struct closely, but the whole network_server.scheduler section is absent from the file — so nothing related to Class-B enqueue timing is exposed there.
- In the upstream open-source project, this delay corresponds to a hardcoded Go constant (ClassBEnqueueMargin, default 5 s) that isn't bound to any config key at all — by design, not configurable even upstream.
- Since Milesight's UG65 build isn't published as source, we can't confirm the exact value you're using (our measurement suggests ~2.05 s, not the 5 s upstream default), or patch/tune it ourselves.
Request: could this margin be exposed as a configurable parameter (via loraserver.json, an env var, or the web UI) in a future firmware update? For low-latency alarm/actuation use cases like ours, being able to tune this down (with the understanding that we take on the risk of missed ping-slots if set too aggressively) would be very valuable — and would save your support team from getting repeat questions like this one.
Happy to share our full measurement logs if useful for reproducing this internally.
Setup for reference: UG65 firmware 60.0.0.50, EU868, Class-B ping-slot periodicity 960 ms (ping_nb=128).
Thanks!
0 Votes
0 Comments
Login or Sign up to post a comment