discv4-crawl

Background

Geth now ships with an implementation of EIP-1459. This EIP defines a way to put devp2p node lists behind a DNS name. There are a couple of things worth knowing about this system:

EIP-1459 is intended to be a replacement for hard-coded bootstrap node lists that we maintain in Ethereum clients.
This is a centralized system where all nodes configured with a certain name resolve subdomains of the name find bootstrap nodes.
The node list is signed with a key which will be hard-coded into the client (i.e. geth) and which we should keep in a secure place.

To create suitable bootstrap node lists for all common networks, we have devised a scheme where software crawls the discovery DHT, then creates a list of all found nodes in JSON format. The crawler software can filter this list and has a built-in deployer that can install the DNS records.

You can read the DNS Discovery Setup Guide for more information about the discovery DHT crawler.

Description

This repository contains the scripts used to automatically generate the list of nodes that are published to the multiple DNS zones. The node list is also automatically pushed to this repository.

Running with docker

Environment variables

Name	Default	Description
CRAWL_GIT_REPO	https://github.com/skylenet/discv4-dns-lists.git	Git repository `used to clone and push the node list
CRAWL_GIT_BRANCH	master	Git branch used for the fetch and push
CRAWL_GIT_PUSH	false	When set to true, it will push the node lists to the git repository
CRAWL_GIT_USER	crawler	Git username. Will appear in the commit messages.
CRAWL_GIT_EMAIL	crawler@localhost	Git email address. Will appear in the `commit messages.
CRAWL_DNS_DOMAIN	nodes.example.local	DNS domain suffix used for the directory structure
CRAWL_TIMEOUT	30m	The time spent crawling the discovery DHT
CRAWL_INTERVAL	300	Interval, in seconds, between multiple executions.
CRAWL_RUN_ONCE	false	Set to true if you only want to run the execution once.
CRAWL_DNS_SIGNING_KEY	/secrets/key.json	Path to the signing key. Won't sign if the file doesn't exist.
CRAWL_DNS_PUBLISH_ROUTE53	false	Publish the TXT records to a DNS zone on AWS Route53
ROUTE53_ZONE_ID	``	Route53 DNS zone identifier. This is the zone where the records will be published to.
AWS_ACCESS_KEY_ID	``	AWS access key
AWS_SECRET_ACCESS_KEY	``	AWS secret access key
CRAWL_DNS_PUBLISH_CLOUDFLARE	false	Publish the TXT records to a DNS zone on Cloudflare
CLOUDFLARE_API_TOKEN	``	API token used for the Cloudflare API
CLOUDFLARE_ZONE_ID	``	Cloudflare DNS zone identifier. This is the zone where the records will be published to.
INFLUXDB_METRICS_ENABLED	false	Set to true if you want to send metrics to InfluxDB
INFLUXDB_URL	http://localhost:8086	Address of the InfluxDB API
INFLUXDB_DB	metrics	Database name
INFLUXDB_USER	user	Username for InfluxDB
INFLUXDB_PASSWORD	password	Password for InfluxDB
PROMETHEUS_METRICS_ENABLED	true	Enable prometheus metrics endpoint
PROMETHEUS_METRICS_LISTEN	0.0.0.0:9100	Server listening

Building the image

$ docker build -t discv4-crawl .

Run examples

Run the list generation and push the results to git via SSH:

$ docker run -it \
    -v "$HOME/.ssh/crawler:/root/.ssh" \  # Needed if you use git via SSH
    -v "$HOME/secrets/secret-signing-key.json:/secrets/key.json" \ # Only needed if you want to sign the node lists
    -e CRAWL_TIMEOUT=10m \ # Specify your custom timeout
    -e CRAWL_GIT_REPO=git@github.com:skylenet/discv4-dns-lists.git \ # Use SSH instead of HTTPS
    -e CRAWL_GIT_PUSH=true \ # Specify that we want to push the changes
    skylenet/discv4-crawl