Livepeer Monitoring Supercontainer

a bundled prometheus/grafana container with a config templating for monitoring livepeer instances in different deployments

Quick Installation

docker pull livepeer/monitoring

Setup

Args

note all Args can be passed as environment variables by adding the prefix LP_ to the arg, for example --nodes would become LP_NODES if passed as an env variable.

• --mode : standalone for local bare metal deployments of livepeer or docker-compose mode for running as a part of a docker-compose manifest or kuberenetes mode for running as a part of a kube pod

• --nodes: a comma separated list of the livepeer nodes and their cli port we'd like to monitor, example: --nodes=localhost:7935,localhost:7936, this isn't required in the kubernetes deployments since discovery is done automatically using the prometheus.io/scrape labels.

• --kube-namespaces : comma separated list of namespaces to monitoring in the kubernetes deployment, this is needed for certain special deployments, it defaults to an empty array.

• --prometheus-kube-scrape: string annotation for scraping a kube pod. Ex. If the value for this flag is scrape_this_pod then all kube pods to be scraped should have the annotation prometheus.io/scrape_this_pod. The value for this flag must follow the Prometheus requirements for naming and match the regex [a-zA-Z_][a-zA-Z0-9_]* (ACII letters, numbers and underscores).

• --pagerduty-service-key: the service key (not API key!) for the Prometheus integration on your Pagerduty service to receive alerts from Prometheus Alertmanager.

ENVs

these env variables cannot be passed as a arg, this is because we use them in the simple docker templating for stuff like grafana datasource

• LP_PROMETHEUS_ENDPOINT : a full url of the prometheus endpoint, this defaults to http://localhost:9090 , but in our kubernetes deployments and docker-compose, this need to be changed

Grafana Envs

All GF_ prefixed envs are passed to grafana , you can find out more details at the official grafana docs

Usage examples

# simple standalone setup example using envs
$ docker run --net=host --env LP_MODE=standalone --env LP_NODES=localhost:9735,localhost:7936 livepeer/monitoring:latest

# using args
$ docker run --net=host livepeer/monitoring:latest --mode standalone --nodes=localhost:9735,localhost:7936

Note: On OSX omit --net=host and use host.docker.internal instead of localhost for the --nodes. Port forwarding using the -p flag is also required. For example:

docker run livepeer/monitoring:latest --mode standalone --nodes=host.docker.internal:7935

The nodes specified must be run with the -monitor flag so that they expose a /metrics endpoint that Prometheus can scrape.

Dashboards are available at the usual 3000 port for Grafana and 9090 for Prometheus. You can change the port used to access the dashboards using the Docker port forwarding -p flag.