exporter

README

fio-prometheus-exporter

This is a simple prometheus exporter for FIO nodeos nodes. It can connect to multiple nodes to display a few critical statistics.

Data Provided

Note: not all of the APIs are enabled by default because they are dangerous, and could allow bad things to happen. If you enable the net or producer API plugins, be sure to block access to /v1/net/* and /v1/producer/* from outside connections!

Labels

By default the following labels are provided:

• chain_id will be 'mainnet', 'testnet', or if not recognized, the hash.
• version the friendly version string provided by the node.
• host the hostname of the instance (provided because the {{instance}} variable will not be the actual host)

Any metrics related to a producer will have an addition label, aptly named producer

Standard endpoints (should work by default)

• fio_info_head_block head block number from get_info endpoint

• fio_info_head_lag seconds behind the head block

• fio_info_lib last irreversible block number from get_info endpoint

• fio_producer_is_active whether producer is set to active

• fio_producer_is_top21 whether producer is in the top 21: 0 if false

• fio_producer_last_bpclaim_delta seconds since last bpclaim

• fio_producer_rank producer rank

• fio_producer_schedule_active active producer schedule

• fio_producer_schedule_pending pending producer schedule

• fio_producer_schedule_proposed proposed producer schedule

• fio_producer_votes producer votes in FIO

• fio_runtime_max_transaction_time maximum transaction time (http_plugin)

Non standard endpoints:

Requires eosio::producer_api_plugin

• fio_producer_producing if producer api is paused: 1 means actively signing blocks, otherwise null

Requires eosio::db_size_api_plugin

• fio_db_free_bytes bytes remaining for memory state
• fio_db_used_bytes bytes used by memory state

Requires eosio::net_api_plugin

• fio_net_connected_peers number of connected peers
• fio_net_disconnected_peers number of unreachable peers
• fio_net_syncing_peers number of peers catching up

Grafana

A sample dashboard is available in the grafana folder in this repository:

Setup the job in the prometheus.yml like this:

scrape_configs:
  - job_name: 'fio'
    static_configs:
      - targets: ["your-node-here:13856"]

Running

Usage of fio-prometheus-exporter:
  -no-db
        do not attempt to use the db API
  -no-net
        do not attempt to use the net API
  -no-producer
        do not attempt to use the producer API
  -p int
        port to listen on (default 13856)
  -u string
        nodes to monitor, comma seperated list of http urls (default "http://localhost:8888")

Multiple hosts can be provided by seperating the urls with a comma, for example:

fio-prometheus-exporter -u http://nodeos1:8888,http://nodeos2:8888,http://nodeos3:8888

Docker

A Dockerfile is available, which will build a trimmed from-scratch container, with the openssl libraries and root certificates needed for connecting to a TLS-enabled host.

# build:
docker build -t fio-prometheus-exporter .
# run
docker run -d --restart unless-stopped -p 13856:13856 --name fio-prometheus-exporter fio-prometheus-exporter -u http://nodeos1:8888,http://nodeos2:8888,http://nodeos3:8888

Documentation

Index

• func Serve()
• type NotMonitoredErr
  • func (nm NotMonitoredErr) Error() string
• type NullErr
  • func (ne NullErr) Error() string

Functions

func Serve

func Serve()

Serve starts the exporter.

Types

type NotMonitoredErr

type NotMonitoredErr struct {
	Api string
}

NotMonitoredErr will be returned if an endpoint is not available, or has been explicitly excluded.

func (NotMonitoredErr) Error

func (nm NotMonitoredErr) Error() string

type NullErr

type NullErr struct{}

NullErr will be returned when a response is empty, primarily when getting information about whether a node has block production enabled.

func (NullErr) Error

func (ne NullErr) Error() string