Chirpstack v4 on Azure Kubernetes in Production

Hello everyone

I am seeking advice and guidance. I am beginner in Chirpstack and trying to setup Chirpstack (along with gateway bridge) in Azure Kubernetes Service for one of our clients. Goal is to setup LoraWAN server for getting data from ~25 gateways spread in 3 - 4 cities.

After going through the forum, I have seen people using Chirpstack in Production environment using both VM as well as contianers on Kubernetes but less is mentioned around the correct setup for Production environment to provide scaling capabilities and High Availability (in both scenarios)

Can you please provide guidance around what should be right set of configuration for deploying Chirpstack in Production on Kubernetes? What things to be mindful about from security standpoint?

@brocaar - I know you have provided some guidance around the stateless vs stateful nature of underlying components but that information is scattered in multiple posts. Please if you could provide your take on above use case.

@bconway - I have seen your posts that you are running Chirpstack on Kubernetes, so your practical guidance around which chirpstack services need to run as stateful sets vs deployments and other architecture pointers would be really appreciated.

Also, this customer would have other types of devices as well using M-Bus, so intent is to get data from both types of devices to common InfluxDB, so anything in Chirpstack component (like common MQTT) that can be utilized for both M-Bus and Lora devices would also be more suited, but if it makes the architecture difficult to manage by newbies then we would want both Lora and M-Bus paths separate and converge at the InfluxDB.

This is not just directed to you both but I have seen you have great knowledge in Chirpstack deployment scenarios, hence, mentioned you both but I seek guidance from everyone who are running Chipstack v4 in Production on cloud.

Apologies for this open ended and very subjective topic but your guidance can really help us moving in right direction.



Please stop filling this forum with ChatGPT responses. It’s very transparent.