Skip to content

Latest commit

 

History

History
183 lines (132 loc) · 8.9 KB

TEST_NETWORK.md

File metadata and controls

183 lines (132 loc) · 8.9 KB
Raw

Network Overview

After we have set up a series of TLS and ECert CA services, we'll use the CAs to generate Local MSP structures for all of the nodes, using the local MSPs to launch our network peers and orderers.

TL/DR :

./network up
...
✅ - Creating local node MSP ...
✅ - Launching orderers ...
✅ - Launching peers ...
🏁 - Network is ready.

Fabric Node MSP Context

Before we launch the network peers and orderers, each node in the network needs to have available:

  • TLS Root Certificates for all organizations in the network
  • TLS Certificates and Signing Keys for SSL server/hostname verification of the network node
  • Enrollment Certificates validating the network node identity (local MSP)
  • Enrollment Certificates for an Admin identity / role for the organization.

In order to create the local node MSP, we must first register and enroll the node identities with the ECert CAs, and then organize the TLS and MSP certificates into a location suitable for launching the network services.

The key steps in this process are:

In the test network, each organization includes a function that wraps the registration, enrollment, and MSP aggregation into a series of fabric-ca-client calls. The script will be executed directly on the org's ECert CA pod, with access to the persistent volume for storage of the MSP and TLS certificates. While this is largely boilerplate scripting, the process is straightforward: For each node in the network, we'll use the CAs to generate TLS+MSP certificates, bundling into an MSP with a config.yaml specifying the fabric roles associated with the target usage in the network.

For example, the ordering organization sets up the node local MSP with:

# Each identity in the network needs a registration and enrollment.
fabric-ca-client register --id.name org0-orderer1 --id.secret ordererpw --id.type orderer --url https://org0-ca --mspdir $FABRIC_CA_CLIENT_HOME/org0-ca/rcaadmin/msp
fabric-ca-client register --id.name org0-orderer2 --id.secret ordererpw --id.type orderer --url https://org0-ca --mspdir $FABRIC_CA_CLIENT_HOME/org0-ca/rcaadmin/msp
fabric-ca-client register --id.name org0-orderer3 --id.secret ordererpw --id.type orderer --url https://org0-ca --mspdir $FABRIC_CA_CLIENT_HOME/org0-ca/rcaadmin/msp

fabric-ca-client enroll --url https://org0-orderer1:ordererpw@org0-ca --csr.hosts org0-orderer1 --mspdir /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer1.org0.example.com/msp
fabric-ca-client enroll --url https://org0-orderer2:ordererpw@org0-ca --csr.hosts org0-orderer2 --mspdir /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer2.org0.example.com/msp
fabric-ca-client enroll --url https://org0-orderer3:ordererpw@org0-ca --csr.hosts org0-orderer3 --mspdir /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer3.org0.example.com/msp

# Create an MSP config.yaml (why is this not generated by the enrollment by fabric-ca-client?)
echo "NodeOUs:
  Enable: true
  ClientOUIdentifier:
    Certificate: cacerts/org0-ca.pem
    OrganizationalUnitIdentifier: client
  PeerOUIdentifier:
    Certificate: cacerts/org0-ca.pem
    OrganizationalUnitIdentifier: peer
  AdminOUIdentifier:
    Certificate: cacerts/org0-ca.pem
    OrganizationalUnitIdentifier: admin
  OrdererOUIdentifier:
    Certificate: cacerts/org0-ca.pem
    OrganizationalUnitIdentifier: orderer" > /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer1.org0.example.com/msp/config.yaml

cp /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer1.org0.example.com/msp/config.yaml /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer2.org0.example.com/msp/config.yaml
cp /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer1.org0.example.com/msp/config.yaml /var/hyperledger/fabric/organizations/ordererOrganizations/org0.example.com/orderers/org0-orderer3.org0.example.com/msp/config.yaml

External Chaincode Builders

Running Fabric in Kubernetes places some unique constraints on the Chaincode lifecycle:

  • Many cloud-native vendors rely on containerd.io to manage the lifecycle of containers within a cluster. By contrast, Fabric assumes the presence of a Docker daemon to compile and launch chaincode containers. Without a local Docker daemon, Fabric's default chaincode pipeline is doomed!

  • For security and operational concerns, it is a "non-starter" to run a docker daemon on Kubernetes worker nodes.

  • For cloud-ready development, test, validation, CI/CD, and production practices, the use of the Chaincode as a Service pattern provides a vastly superior user experience.

  • Running Chaincode builds in Docker in Docker, running in Kubernetes in Docker is ... interesting. Let's step back and keep it simple.

In the Kubernetes Test Network, we've incorporated the default ccaas external builder (See fabric #2884) as an accelerator for working with Chaincode-as-a-Service on Kubernetes. For ccaas smart contracts, when chaincode is installed on a peer, the external builder binaries will be invoked, bypassing the reliance on a local Docker daemon running in Kubernetes.

This configuration is accomplished by registering an external builder in the peer core.yaml:

    externalBuilders:
      - name: ccaas_builder
        path: /opt/hyperledger/ccaas_builder
        propagateEnvironment:
          - CHAINCODE_AS_A_SERVICE_BUILDER_CONFIG

To trigger the external builder for a chaincode service, set the metadata.json type attribute to ccaas. E.g.:

{
  "type": "ccaas",
  "label": "basic_1.0"
}
  • Pro tip: Use the companion container registry at localhost:5000 to deploy custom chaincode into the test network.
  • Pro tip: Deploy a chaincode with address: host.docker.internal:9999 and attach your chaincode in a debugger.

Starting Peers and Orderers

✅ - Launching orderers ...
✅ - Launching peers ...

Once the local MSP structures for the network nodes have been created, the orderers and peers may be launched in the namespace. System nodes will read base configuration files (orderer.yaml and core.yaml) from the organization config folder, made available in Kubernetes as the ${org}-config config map.

Each orderer and peer creates one Deployment, Pod, and Service in the namespace. In addition, each org defines an orgN-peerM-config ConfigMap with environment variable overrides replacing the default settings in the core.yaml file. Note that each node's environment includes pointers to the node local MSP folders, certificates, and TLS signing keys that we generated above.

Note that the deployment yaml files include some basic template substitution and parameters. For simplicity and clarity, we elected to use basic string substitution with envsubst, rather than introduce a Kube template binding system (e.g. Helm, Kustomize, Kapitan, Ansible, etc.) for manipulating yaml templates. Any environment variables declared in the ./network script will be applied to these k8s templates. E.g.:

export FABRIC_VERSION=2.4.2
...

cat kube/org0/org0-orderer1.yaml | envsubst | kubectl -n $NS -f -
cat kube/org0/org0-orderer2.yaml | envsubst | kubectl -n $NS -f -
cat kube/org0/org0-orderer3.yaml | envsubst | kubectl -n $NS -f -

# Wait for the orderers to completely start before launching the network peer nodes.
kubectl -n $NS rollout status deploy/org0-orderer1
kubectl -n $NS rollout status deploy/org0-orderer2
kubectl -n $NS rollout status deploy/org0-orderer3

Next Steps :

After the peers and orderers have started, the Kube namespace includes pods, deployments, and service bindings for:

  • Org0 (org0.example.com):

    • ECert Certificate Authority : https://org0-ca
    • Orderer1 : grpcs://org0-orderer1
    • Orderer2 : grpcs://org0-orderer2
    • Orderer3 : grpcs://org0-orderer3

  • Org1 (org1.example.com):

    • ECert Certificate Authority : https://org1-ca
    • Peer Node 1 : grpcs://org1-peer1
    • Peer Node 2 : grpcs://org1-peer2

  • Org2 (org2.example.com):

    • ECert Certificate Authority : https://org2-ca
    • Peer Node 1 : grpcs://org2-peer1
    • Peer Node 2 : grpcs://org2-peer2

Next : Working With Channels