Telemetry

[JoblersTune]

Aim

• Provide telemetry for the network
• Enabled by default

Desired metrics

• Total money [per time frame] taken from packet data
• Average money [per time frame] taken from packet data
• Amount of transactions taken from outgoing payments that have been at least partially successful (i.e. some value in the send amount)

Basic setup

• Rafiki will include the OpenTelemetry SDK to generate metrics, and contextual information
• Using the sidecar pattern with an OpenTelemetry Collector to gather metrics
• Will need the option to enable/disable metric collection
• This functionality will be included within the Rafiki repo
• Using Prometheus for storing, and querying metrics (using the counter primitive)
• Using Grafana for creating dashboards and visualizing metrics
• Running Prometheus and Grafana in containers managed on AWS

Integration

• Rafiki code base needs to expose the desired metrics as HTTP endpoints
• Configure the OpenTelemetry Collector to gather the metrics from Rafiki
• Configure Prometheus to scrape metrics from the OpenTelemetry Collector (uses a pull-based approach)
• In Grafana, you'll configure Prometheus as a data source (can use PromQL to write queries against the data collected by Prometheus.)

Linked Issues:

• [Telemetry] Set up AWS Prometheus & Grafana #1956
• [Telemetry] - Instrument Rafiki #1913
• [Telemetry] - Set up Grafana to view metrics #1923
• [Telemetry] - update Rafiki README with telemetry disclaimer #1924

[JoblersTune]

AWS access key and secret for the open collector user are stored in 1Password