ChirpStack v4.0.0-rc.2
After many months of development and testing, we are really excited to share ChirpStack v4.
The aim of ChirpStack v4 is to make it significantly easier to setup and use ChirpStack, compared to the previous version/ One of the major changes that you will notice is that the ChirpStack Network Server and ChirpStack Application Server have been merged into a single component. Over the years we have seen many issues reported on the forum and GitHub, related to setting up and connecting both services.
ChirpStack v4 also provides multi-region support out-of-the-box, including region configuration. No longer it is needed to define your own configuration file or setup multiple ChirpStack Network Server instances to serve multiple regions simultaneously.
A big thank you to the ChirpStack community for supporting and contributing to the ChirpStack project! Please find below a breakdown of all the new features and changes that v4 brings.
Main features and changes
Multi-region support
ChirpStack v4 adds multi-region support, removing the need to setup multiple ChirpStack Network Server instances. Configuration files are included for the common regions (as defined by the LoRa Alliance), which should help getting started with ChirpStack.
Each enabled region has its own gateway backend, making it possible to use one or multiple MQTT brokers for the different gateway pools. In case a single MQTT broker is used, each backend must be configured with its own MQTT topic prefix (e.g. eu868/gateway/...
).
ChirpStack v4 also supports multiple configurations of the same region, e.g. to configure a US915 for 8 channels and to configure a US915 band for 16 channels.
TTN device repository support
ChirpStack v4 adds support for importing the TTN LoRaWAN Devices repository as device-profile templates, including codec functions if these are defined in this repository.
Device metrics dashboard
In the device-profile template and / or device-profile, it is possible to define the measurements that are exposed by the device in the decoded payload. Once defined, ChirpStack will automatically aggregate and store this data. These metrics can be viewed in the web-interface on the device dashboard.
Configuration
Directory
Instead of using a single configuration file (e.g. chirpstack-network-server.toml
), ChirpStack makes use of a configuration directory such that the configuration can be split in multiple files. By default you will find a single chirpstack.toml
configuration file, and many region_...toml
configuration files, split by region.
Environment variables
ChirpStack v4 removes the TOML hierarchy to environment variable mapping. Instead it allows you to define the variables like $MY_CONFIGURATION
, which will get automatically substituted when an environment variable is found with the name MY_CONFIGURATION
.
API
The REST interface that was present in ChirpStack Application Server v3 has been removed, in favor of the gRPC API interface (please see the api/
folder of the repository for the API definitions). However, a gRPC to REST interface bridge component will be provided as a separate service. Please note that
in v3, this bridge component was embedded and REST interface calls were internally translated to gRPC calls. Therefore, gRPC was always recommended interface to use.
The ChirpStack v4 gRPC API does support server reflection, making it possible to use for example gRPC UI or BloomRPC as development interface.
ChirpStack Gateway Bridge v3 compatibility
ChirpStack v4 is fully compatibility with the latest version of ChirpStack Gateway Bridge v3. This should help migrating from v3 to v4. Please note that the ChirpStack Gateway Bridge must be configured with the protobuf
marshaler.
UI rewrite
ChirpStack v4 contains a rewrite of the ChirpStack Application Server v3 UI. The new UI aims to be more user-friendly. Under the hood the API interface has been ported to gRPC-web and all code has been ported to Typescript.
Other changes
Passive Roaming improvements
The implementation of Passive Roaming has been improved, adding support for appending /sns
and /fns
server endpoint suffixes. The usage of this suffix is not specified in the Backend Interfaces specification, but is required by some other network-server implementations.
UUID identifiers
All identifiers that are exposed have been changed to UUID. Previously most identifiers (e.g. users, applications…) were incremental integers. In case ChirpStack is setup as multi-tenant instance, this could expose some information about the number of clients on the network. The migration script (see below)
will migrate these integers by converting these as strings, prefixed with zeros in the UUID format. E.g. ID 123
would be converted to the UUID string 00000000-0000-0000-0000-000000000123
.
String identifiers in API
All binary identifiers have been changed to string type in the API. While binary fields are more efficient, these were confusing when encoded as JSON as the Protobuf to JSON mapping uses base64 encoding for binary fields. For example, a Gateway ID 0102030405060708
was encoded as AQIDBAUGBwg=
in JSON.
JavaScript codec engine
The JavaScript codec engine is based on QuickJS, which is an embeddable JavaScript engine which supports the ES2020 specification. In the near future support for the NodeJS Buffer class will be added.
API interface changes
While the structure of API messages is roughly the same as the ChirpStack Application Server API interface, some small changes have been made.
Integration events
The integration event messages have been restructured for better consistency. Each event message has a deviceInfo
field which holds device-related information (tenant id & name, application id & name, device-profile id & name, device EUI & name and tags).
Development changes
Single repository
ChirpStack v4 will make it a lot easier to make customizations, especially when API changes are involved, as API definitions are no longer separated from the code. In v3 these definitions were moved to an external repository to avoid cross dependencies.
Rust
For ChirpStack v4, it was decided to use Rust rather than Go. This was not an easy choice and the arguments for this decision are debatable. However, as most code was touched during the ChirpStack Application Server and ChirpStack Network Server merge, it was the only moment to re-consider this. The Rust memory management prevents many memory related pitfalls and helps catching bugs at compile time rather than runtime.
Migrating from v3 to v4
The recommended way to migrate from v3 to ChirpStack v4 is to create a new PostgreSQL and Redis database and to use the ChirpStack v3 to v4
migration script. This script will copy all the data from the “old” into the “new” database. While the script does not make any modifications to the old database, it is always recommended to make a backup first.
In a few minutes the www.chirpstack.io documentation will be updated with the ChirpStack v4 documentation (+ link to ChirpStack v3 documentation).
Note: As no stable v4.0.0 release exists at this point, the release-candidates will be pushed to both the stable
and testing
v4 respository channels (Debian / Ubuntu). After the first stable v4.0.0 release, only the testing
channel will be used for test-releases.