Skip to content

Latest commit

 

History

History
136 lines (119 loc) · 9.69 KB

README.md

File metadata and controls

136 lines (119 loc) · 9.69 KB
Raw

fabric-operation

This project contains scripts that let you define, create, and test a Hyperledger Fabric network in Kubernetes locally or in a cloud. Supported cloud services include AWS, Azure, and Google Cloud. The fabric network parameters can be specified by a property file, such as the sample network, netop1.env.

The scripts support both docker-compose and kubernetes. All steps are done in docker containers, and thus you can get a Fabric network running without pre-downloading any artifact of Hyperledger Fabric.

This utility is implemented using bash scripts, and thus it does not depend on any other scripting tool or framework, either.

Prerequisites

  • Your workstation must support bash shell scripts.
  • If you want to create and test a Fabric network on local host, you need to install docker-compose and/or kubernetes locally, i.e.,
    • Install Docker and Docker Compose as described here.
    • Mac user can enable kubernetes as described here.
    • I have not tested the scripts with Minikube, although I would expect it to work without much change.
  • If you want to create and test a Fabric network in a cloud, you would not need to download anything except a CLI required to access the corresponding cloud service. We currently support Amazon EKS, Azure AKS, and Google GKE. Other cloud services may be supported in the future.
    • For AWS, refer the scripts and instructions in the aws folder.
    • For Azure, refer the scripts and instructions in the az folder.
    • For Google cloud, refer the scripts and instructions in the gcp folder

Prepare Kubernetes namespace

This step is necessary only if you use Kubernetes. So, skip it when docker-compose is used.

cd ./namespace
./k8s-namespace.sh create

This command creates a namespace for the default Fabric operator company, netop1. It also sets netop1 as the default namespace, so you won't have to specify the namespace in the following kubectl commands.

To revoke to the default namespace for docker-desktop, you can use the following command:

kubectl config use-context docker-desktop

Start CA server and generate crypto data

Following steps use docker-desktop Kubernetes on Mac to start fabric-ca PODs and generate crypto data required by the sample network, netop1.

cd ../ca
# cleanup old ca-server data
rm -R ../netop1.com/canet
./ca-server.sh start
# wait until the 3 PODs for ca server and client are in running state
./ca-crypto.sh bootstrap

You can edit the network specification netop1.env if you want to use a different operating company name, or make it run more orderer or peer nodes. The generated crypto data will be stored in the folder netop1.com on localhost, or in a cloud file system, such as Amazon EFS, or Azure Files.

These scripts take 2 additional parameters, e.g.,

./ca-server.sh start -p <config_file> -t <env_type>
./ca-crypto.sh bootstrap -p <config_file> -t <env_type>

where config_file is file in the config folder with a suffix .env that contains the fabric network specification; env_type can be k8s, docker, aws, or az. When no parameter is specified, it uses default -p netop1 -t k8s. Refer ca folder for more detailed description of these scripts.

  • k8s uses the local docker-desktop kubernetes on Mac. Non-Mac users may use docker option below, or try Minikube (which has not been tested).
  • docker uses docker-compose.
  • aws uses AWS EKS when executed on a bastion host of an EC2 instance. Refer the folder aws for more details on AWS.
  • az uses Azure AKS when executed on a bastion VM instance in Azure. Refer the folder az for more details on Azure. More cloud support will be added in the future.
  • gcp uses Google GKE when executed on a bastion host in Google Cloud. Refer the folder gcp for more details on Google Cloud.

Sample crypto data

When the above steps are executed on localhost, the crypto data will be stored in netop1.com, which is specified by FABRIC_ORG in the network definition file netop1.env. The resulting crypto data is similar to that generated by the fabric cryptogen tool as demonstrated by fabric-samples. However, by using a fabric CA server in the above step, the generated certificates will include a few extra attributes that would make them usable for cloud deployment using kubernetes, as well as attribute-based-access-control (ABAC). Besides, CA server is also more flexible for generating certificates for more nodes and users in production environment as the network grows. Although the CA servers use a self-signed root CA for simplicity, you may supply your real root CA for production deployment.

You may verify the generated crypto data by using a preconfigured sample network as described in docker-netop1. However, if you do not have a local hyperledger fabric environment, you can skip the test and read on. The following steps will show you how to start a fabric network by using a few simple scripts even if you do not have a fabric development environment.

Generate MSP definition and genesis block

The following script generates a genesis block for the sample network in Kubernetes using 2 peers and 3 orderers with etcd raft consensus.

cd ../msp
./msp-util.sh start
# wait until the too POD is running
./msp-util.sh bootstrap

It also generates transactions for creating a test channel, mychannel, for smoke testing. Similar to other scripts, this command also accepts 2 parameters, e.g.,

./msp-util.sh start -p <config_file> -t <env_type>
./msp-util.sh bootstrap -p <config_file> -t <env_type>

so you can specify a different network definition file, or generate artifacts for other deployment environment, e.g., docker, aws, az, or gcp. Refer msp folder for more detailed description of these scripts.

Start and smoke test the Fabric network

The following script will start and test the sample fabric network by using the docker-desktop Kubernetes on a Mac:

cd ./network
./network.sh start
# wait until 3 orderer and 2 peer nodes are running, Raft leader is elected in orderers
./network.sh test
./network.sh shutdown

After the network startup, use kubectl logs orderer-2 to check RAFT leader election result. When RAFT leader is elected, the log should show

INFO 101 Raft leader changed: 0 -> 2 channel=netop1-channel node=2

Before you shutdown the network, you can verify the running fabric containers by using kubectl, e.g.,

kubectl get pod,svc -n netop1

Note that the scripts use the operating company name netop1, as a Kubernetes namespace, and so they can support multiple member organizations.

After the smoke test succeeds, you should see a test result of 90 printed on the screen. If you used docker-compose for this excersize (as described below), you can look at the blockchain state via the CouchDB futon UI at http://localhost:7056/_utils, which is exposed for docker-compose only because it is not recommended to expose CouchDB in production configuration using Kubernetes.

Start gateway service and use REST APIs to test chaincode

Refer gateway for more details on how to build and start a REST API service for applications to interact with one or more Fabric networks. The following commands will start a gateway service that exposes a Swagger-UI at http://localhost:30081/swagger.

cd ../service
./gateway.sh start

Operations for managing the Fabric network

The above bootstrap network is for a single operating company to start a Fabric network with its own orderer and peer nodes of pre-configured size. A network in production will need to scale up and let more organizations to join and co-operate. Organizations may create their own Kubernetes networks using the same or different cloud service providers. We provide scripts to support such network activities.

The currently supported operations include

  • Create and join new channel;
  • Install and instantiate new chaincode;
  • Add new peer nodes of the same bootstrap org;
  • Add new orderer nodes of the same bootstrap org;
  • Add new peer org to the same Kubernetes cluster;

Refer operations for description of these activities. More operations (as described in TODO bellow) will be supported in the future.

Non-Mac users

If you are not using a Mac, you can run these scripts using docker-compose, Amazon EKS, Azure AKS, or Google GKE. Simply add a corresponding env_type in all the commands, e.g.,

  • ./ca-server.sh start -t docker to use docker-composer, or
  • ./ca-server.sh start -t aws to use AWS as described in the folder aws, or
  • ./ca-server.sh start -t az to use Azure as described in the folder az, or
  • ./ca-server.sh start -t gcp to use Google Cloud as described in the folder gcp, or
  • try to verify if the scripts would work on Minikube.

When docker-compose is used, start and test the Fabric network using the following commands:

cd ./network
./network.sh start -t docker
./network.sh test -t docker
./network.sh shutdown -t docker

TODO

Stay tuned for more updates on the following items:

  • Add new orderer org to the same bootstrap Kubernetes cluster for etcd raft consensus;
  • Add new orderer org to a new Kubernetes cluster;
  • Add new peer org to a new Kubernetes cluster;
  • Test multiple org multiple Kubernetes clusters across multiple cloud providers.