Private network

This page describes how to set up a local cluster of nodes, advises how to make them private, and how to hook up your nodes on the eth-netstat network monitoring app. A fully controlled Ethereum network is useful as a backend for network integration testing (core developers working on issues related to networking/blockchain synching/message propagation, or dApp developers testing multi-block and multi-user scenarios).

This guide assumes that you're already able to build geth by following the build instructions

Setting up multiple nodes

In order to locally run multiple Ethereum nodes, you have to make sure:

• Each instance has a separate data directory (--datadir)

• Each instance runs on a different port (both ETH and rpc) (--port and --rpcport)

• In case of a cluster, the instances must know about each other

• The ipc endpoint is unique or the ipc interface is disabled (--ipcpath or --ipcdisable)

To start the first node, let's make port explicit and disable ipc interface:

geth --datadir="/tmp/eth/60/01" -verbosity 6 --ipcdisable --port 20630 --rpcport 8101 console 2>> /tmp/eth/60/01.log

We started the node with the console in order to grab the enode url - for instance:

> admin.nodeInfo.enode
enode://8c544b4a07da02a9ee024def6f3ba24b2747272b64e16ec5dd6b17b55992f8980b77938155169d9d33807e501729ecb42f5c0a61018898c32799ced152e9f0d7@9[::]:20630

[::] will be parsed as localhost (127.0.0.1). If your nodes are on a local network, check each individual host machine and find your ip with ifconfig (on Linux and MacOS):

$ ifconfig|grep netmask|awk '{print $2}'
127.0.0.1
192.168.1.97

If your peers are not on the local network, you'll need to know your external IP address (use a service) to construct the enode url.

Now you can launch a second node with:

geth --datadir="/tmp/eth/60/02" --verbosity 6 --ipcdisable --port 30302 --rpcport 8102 console 2>> /tmp/eth/60/02.log 

If you want to connect this instance to the previously started node, you can add it as a peer from the console with admin.addPeer(enodeUrlOfFirstInstance).

You can test the connection by typing in geth console:

> net.listening
true
> net.peerCount 
1
> admin.peers
...

Local Cluster

As an extension of the above, you can easily spawn a local cluster of nodes. It can also be scripted, including account creation, which is needed for mining. See gethcluster.sh script and the README for usage and examples.

Private network

See [[the Private Network Page|Private network]] for more information.

Setup bootnode

The first time a node connects to the network, it uses one of the predefined bootnodes. Through these bootnodes, a node can join the network and locate other nodes. In the case of a private cluster, these predefined bootnodes are not of much use. Therefore, Elastos.ELA.SideChain.ESC offers a bootnode implementation that can be configured and run on your private network.

It can be run through the command:

> bootnode
Fatal: Use -nodekey or -nodekeyhex to specify a private key

As displayed, the bootnode asks for a key. Each Ethereum node, including bootnodes, is identified by an enode identifier, which is derived from a key. Therefore, you'll need to give the bootnode this key. Since we currently don't have one, we can instruct the bootnode to generate a key (and store it in a file) before it starts.

> bootnode -genkey bootnode.key
I0216 09:53:08.076155 p2p/discover/udp.go:227] Listening, enode://890b6b5367ef6072455fedbd7a24ebac239d442b18c5ab9d26f58a349dad35ee5783a0dd543e4f454fed22db9772efe28a3ed6f21e75674ef6203e47803da682@[::]:20630

(exit with CTRL-C)

The stored key can be seen with:

> cat bootnode.key
dc90f8f7324f1cc7ba52c4077721c939f98a628ed17e51266d01c9cd0294033a

To instruct geth nodes to use our own bootnode(s), use the --bootnodes flag. This is a comma separated list of bootnode enode identifiers.

geth --bootnodes "enode://890b6b5367ef6072455fedbd7a24ebac239d442b18c5ab9d26f58a349dad35ee5783a0dd543e4f454fed22db9772efe28a3ed6f21e75674ef6203e47803da682@[::]:20630"

(what [::] means is explained previously)

Since it's convenient to start the bootnode each time with the same enode, we can give the bootnode program the recently generated key the next time it's started.

bootnode -nodekey bootnode.key
I0216 10:01:19.125600 p2p/discover/udp.go:227] Listening, enode://890b6b5367ef6072455fedbd7a24ebac239d442b18c5ab9d26f58a349dad35ee5783a0dd543e4f454fed22db9772efe28a3ed6f21e75674ef6203e47803da682@[::]:20630

or

bootnode -nodekeyhex dc90f8f7324f1cc7ba52c4077721c939f98a628ed17e51266d01c9cd0294033a
I0216 10:01:40.094089 p2p/discover/udp.go:227] Listening, enode://890b6b5367ef6072455fedbd7a24ebac239d442b18c5ab9d26f58a349dad35ee5783a0dd543e4f454fed22db9772efe28a3ed6f21e75674ef6203e47803da682@[::]:20630

Monitoring your nodes

This page or this README describes how you set up your own monitoring service for a (private or public) local cluster.