Setting Up a Network

This document covers setting up a network on your local machine for various development and testing activities. Unless you are intending to contribute to the development of the Hyperledger Fabric project, you’ll probably want to follow the more commonly used approach below - leveraging published Docker images for the various Hyperledger Fabric components, directly. Otherwise, skip down to the secondary approach below.

Leveraging published Docker images

This approach simply leverages the Docker images that the Hyperledger Fabric project publishes to DockerHub and either Docker commands or Docker Compose descriptions of the network one wishes to create.

Installing Docker

Note: When running Docker natively on Mac and Windows, there is no IP forwarding support available. Hence, running more than one fabric-peer image is not advised because you do not want to have multiple processes binding to the same port. For most application and chaincode development/testing running with a single fabric peer should not be an issue unless you are interested in performance and resilience testing the fabric’s capabilities, such as consensus. For more advanced testing, we strongly recommend using the fabric’s Vagrant development environment.

With this approach, there are multiple choices as to how to run Docker: using Docker Toolbox or one of the new native Docker runtime environments for Mac OSX or Windows. There are some subtle differences between how Docker runs natively on Mac and Windows versus in a virtualized context on Linux. We’ll call those out where appropriate below, when we get to the point of actually running the various components.

Pulling the images from DockerHub

Once you have Docker (1.11 or greater) installed and running, prior to starting any of the fabric components, you will need to first pull the fabric images from DockerHub.

  docker pull hyperledger/fabric-peer:latest
  docker pull hyperledger/fabric-membersrvc:latest

Building your own images

Note: This approach is not necessarily recommended for most users. If you have pulled images from DockerHub as described in the previous section, you may proceed to the next step.

The second approach would be to leverage the development environment setup (which we will assume you have already established) to build and deploy your own binaries and/or Docker images from a clone of the hyperledger/fabric GitHub repository. This approach is suitable for developers that might wish to contribute directly to the Hyperledger Fabric project, or that wish to deploy from a fork of the Hyperledger code base.

The following commands should be run from within the Vagrant environment described in Setting Up Development Environment.

To create the Docker image for the hyperledger/fabric-peer:

cd $GOPATH/src/
make peer-image

To create the Docker image for the hyperledger/fabric-membersrvc:

make membersrvc-image

Starting up validating peers

Check the available images again with docker images. You should see hyperledger/fabric-peer and hyperledger/fabric-membersrvc images. For example,

$ docker images
REPOSITORY                      TAG                 IMAGE ID            CREATED             SIZE
hyperledger/fabric-membersrvc   latest              7d5f6e0bcfac        12 days ago         1.439 GB
hyperledger/fabric-peer         latest              82ef20d7507c        12 days ago         1.445 GB

If you don’t see these, go back to the previous step.

With the relevant Docker images in hand, we can start running the peer and membersrvc services.

Determine value for CORE_VM_ENDPOINT variable

Next, we need to determine the address of your docker daemon for the CORE_VM_ENDPOINT. If you are working within the Vagrant development environment, or a Docker Toolbox environment, you can determine this with the ip add command. For example,

$ ip add

<<< detail removed >>>

3: docker0: <NO-CARRIER,BROADCAST,MULTICAST,UP> mtu 1500 qdisc noqueue state DOWN group default
    link/ether 02:42:ad:be:70:cb brd ff:ff:ff:ff:ff:ff
    inet scope global docker0
       valid_lft forever preferred_lft forever
    inet6 fe80::42:adff:febe:70cb/64 scope link
       valid_lft forever preferred_lft forever

Your output might contain something like inet scope global docker0. That means the docker0 interface is on IP address Use that IP address for the CORE_VM_ENDPOINT option. For more information on the environment variables, see core.yaml configuration file in the fabric repository.

If you are using the native Docker for Mac or Windows, the value for CORE_VM_ENDPOINT should be set to unix:///var/run/docker.sock. [TODO] double check this. I believe that also works.

Assigning a value for CORE_PEER_ID

The ID value of CORE_PEER_ID must be unique for each validating peer, and it must be a lowercase string. We often use a convention of naming the validating peers vpN where N is an integer starting with 0 for the root node and incrementing N by 1 for each additional peer node started. e.g. vp0, vp1, vp2, …


By default, we are using a consensus plugin called NOOPS, which doesn’t really do consensus. If you are running a single peer node, running anything other than NOOPS makes little sense. If you want to use some other consensus plugin in the context of multiple peer nodes, please see the Using a Consensus Plugin section, below.

Docker Compose

We’ll be using Docker Compose to launch our various Fabric component containers, as this is the simplest approach. You should have it installed from the initial setup steps. Installing Docker Toolbox or any of the native Docker runtimes should have installed Compose.

Start up a validating peer:

Let’s launch the first validating peer (the root node). We’ll set CORE_PEER_ID to vp0 and CORE_VM_ENDPOINT as above. Here’s the docker-compose.yml for launching a single container within the Vagrant development environment:

  image: hyperledger/fabric-peer
    - CORE_PEER_ID=vp0
  command: peer node start

You can launch this Compose file as follows, from the same directory as the docker-compose.yml file:

$ docker-compose up

Here’s the corresponding Docker command:

$ docker run --rm -it -e CORE_VM_ENDPOINT= -e CORE_LOGGING_LEVEL=DEBUG -e CORE_PEER_ID=vp0 -e CORE_PEER_ADDRESSAUTODETECT=true hyperledger/fabric-peer peer node start

If you are running Docker for Mac or Windows, we’ll need to explicitly map the ports, and we will need a different value for CORE_VM_ENDPOINT as we discussed above.

Here’s the docker-compose.yml for Docker on Mac or Windows:

  image: hyperledger/fabric-peer
    - "5000:5000"
    - "30303:30303"
    - "30304:30304"
    - CORE_VM_ENDPOINT=unix:///var/run/docker.sock
  command: peer node start

This single peer configuration, running the NOOPS ‘consensus’ plugin, should satisfy many development/test scenarios. NOOPS is not really providing consensus, it is essentially a no-op that simulates consensus. For instance, if you are simply developing and testing chaincode; this should be adequate unless your chaincode is leveraging membership services for identity, access control, confidentiality and privacy.

Running with the CA

If you want to take advantage of security (authentication and authorization), privacy and confidentiality, then you’ll need to run the Fabric’s certificate authority (CA). Please refer to the CA Setup instructions.

Start up additional validating peers:

Following the pattern we established above we’ll use vp1 as the ID for the second validating peer. If using Docker Compose, we can simply link the two peer nodes. Here’s the docker-compse.yml for a Vagrant environment with two peer nodes - vp0 and vp1:

  image: hyperledger/fabric-peer
  command: peer node start
    service: vp0
    - CORE_PEER_ID=vp1
    - vp0

If we wanted to use the docker command line to launch another peer, we need to get the IP address of the first validating peer, which will act as the root node to which the new peer(s) will connect. The address is printed out on the terminal window of the first peer (e.g. and should be passed in with the CORE_PEER_DISCOVERY_ROOTNODE environment variable.

docker run --rm -it -e CORE_VM_ENDPOINT= -e CORE_PEER_ID=vp1 -e CORE_PEER_ADDRESSAUTODETECT=true -e CORE_PEER_DISCOVERY_ROOTNODE= hyperledger/fabric-peer peer node start

Using a Consensus Plugin

A consensus plugin might require some specific configuration that you need to set up. For example, to use the Practical Byzantine Fault Tolerant (PBFT) consensus plugin provided as part of the fabric, perform the following configuration:

  1. In core.yaml, set the peer.validator.consensus value to pbft
  2. In core.yaml, make sure the is set sequentially as vpN where N is an integer that starts from 0 and goes to N-1. For example, with 4 validating peers, set the tovp0, vp1, vp2, vp3.
  3. In consensus/pbft/config.yaml, set the general.mode value to batch and the general.N value to the number of validating peers on the network, also set general.batchsize to the number of transactions per batch.
  4. In consensus/pbft/config.yaml, optionally set timer values for the batch period (general.timeout.batch), the acceptable delay between request and execution (general.timeout.request), and for view-change (general.timeout.viewchange)

See core.yaml and consensus/pbft/config.yaml for more detail.

All of these setting may be overridden via the command line environment variables, e.g. CORE_PEER_VALIDATOR_CONSENSUS_PLUGIN=pbft or CORE_PBFT_GENERAL_MODE=batch

Logging control

See Logging Control for information on controlling logging output from the peer and deployed chaincodes.