Multibuy Service

In the Helium Packet Router architecture, there are multiple load balanced HPRs per region (Europe, Asia, etc). This presents the chance that two geographically close Hotspots in one region actually report to different instances of HPR. Through this double connection, HPRs alone cannot determine whether a packet has already been purchased by the network already.

e.g. One Hotspot reports to HPR A, One Hotspot reports to HPR B. The 'multibuy' is set to 'one', the network may incorrectly transmit two packet reports to the LNS (and packet verifier).

Multi-Buy service fixes this non-communication by allowing HPRs to communicate within a region in order to limit the number of packets purchased by the network (based on multi-buy preference).

As packets come in to HPR A and HPR B, they will check in with Multi-Buy service to ensure the total requested packets is not exceeded.

Features

• Distributed packet counter across load-balanced HPR instances
• Prometheus metrics endpoint
• Automatic cache cleanup (configurable, default 30 minutes)
• Graceful shutdown via SIGTERM/SIGINT

Diagram

flowchart LR
    LNS[LoRaWAN Network Server]
		HPRA & HPRB --> LNS
    subgraph Region
			direction LR
			Multibuy(Multibuy Service)
			Reports[(Reports)]
			HPRA[HPR A] & HPRB[HPR B]
      subgraph Hotspots
        HS1((HS1)) & HS2((HS2)) & HS3((HS3)) & HS4((HS4)) & HS5((HS5)) & HS6((HS6)) & HS7((HS7))
      end
			Sensor{{Sensor X}}
      HPRA ~~~ Multibuy ~~~ HPRB
    end

		HPRA & HPRB --> Multibuy & Reports
		HS1 & HS2 & HS4 & HS7 --> HPRA
		HS3 & HS5 & HS6 --> HPRB
		Sensor -.-> HS1 & HS5

Loading

Building

cargo build --release

Running

# With config file
multi_buy_service -c settings.toml server

# Or with environment variables only
MB__GRPC_LISTEN=0.0.0.0:6080 multi_buy_service server

Testing

cargo nextest run

Docker

docker build -t multibuy-service .

docker run -p 6080:6080 -p 19011:19011 multibuy-service

Configuration

All settings can be configured via a TOML file or environment variables prefixed with MB__ (double-underscore separator).

# log settings for the application (RUST_LOG format)
# Env: MB__LOG
log = "INFO"

# Listen address for gRPC requests
# Env: MB__GRPC_LISTEN
grpc_listen = "0.0.0.0:6080"

# Prometheus metrics endpoint
[metrics]
# Env: MB__METRICS__ENDPOINT
endpoint = "0.0.0.0:19011"

# Cache cleanup interval (humantime format)
# Env: MB__CLEANUP_TIMEOUT
# cleanup_timeout = "30 minutes"

Environment Variables

Variable	Description	Default
MB__LOG	RUST_LOG format log level	INFO
MB__GRPC_LISTEN	gRPC listen address	0.0.0.0:6080
MB__METRICS__ENDPOINT	Prometheus metrics listen address	0.0.0.0:19011
MB__CLEANUP_TIMEOUT	Cache cleanup interval	30 minutes

Metrics

Metric	Type	Description
multi_buy_hit_total	Counter	Total inc requests received
multi_buy_cache_size	Gauge	Number of entries in the cache
multi_buy_cache_cleaned_total	Counter	Total entries removed by cache cleanup

Metrics are exposed at http://{endpoint}/metrics in Prometheus format.