Cello Ansible agent how-to

Cello Ansible agent is a set of Ansible playbooks which allows developers and Fabric network operators to stand up a Hyperledger Fabric network very quickly in a set of virtual or physical servers in cloud or lab.

The following is the list of the general steps you can follow to use the Ansible agent to stand up your own fabric network:

  1. Setup Ansible controller
  2. Create configuration files
  3. Provision the servers
  4. Initialize the servers
  5. Setup fabric network
  6. Verify fabric network

This document also provides the following sections to help you use the Cello Ansible agent:

Cloud configuration file details
Fabric configuration file details
Running an Ansible playbook
Use ssh-agent to help Ansible
Convenient configurations and commands
Use the existing servers
Security rule references

Set up the Ansible controller

You need an Ansible controller to run Ansible playbooks. An Ansible controller can be any machine (VirtualBox VM, your laptop, AWS instance, etc) that has Ansible 2.3.0.0 or greater installed on it.

  1. Install Ansible
  2. Generate a ssh key pair
  3. Clone Ansible agent code
  4. Install Ansible cloud module

Install Ansible

Please follow the offical Ansible installation instructions from the Ansible web site to install Ansible. If you want to start quickly and already have a clean Ubuntu 16.04 system, here are the commands to install Ansible:

    sudo apt-get update
    sudo apt-get install python-pip -y
    sudo pip install 'ansible>=2.3.0.0'

Once it is installed, you should run following command to check its version:

    ansible --version

The results should show something at or above 2.3.0.0. It is entirely possible that you may have to install additional software that Ansible depends on in some older versions of operating systems.

Generate a ssh key pair

Ansible relies heavily on ssh to work with virtual or physical servers. To establish an ssh connection to the servers, you will need an ssh key pair. You may choose to use your own preexisting pair, but it is highly recommended to create a new one for test purposes. Here are the steps to generate a key pair:

    mkdir -p ~/.ssh && cd ~/.ssh && ssh-keygen -t rsa -f fd -P ""

This will create a ssh key pair without a passphrase in your ~/.ssh directory. The name of the private and public key files will be fd and fd.pub, respectively. In the default Ansible agent playbook configuration files, the ssh key is assumed to exist at that location with those names. If you use different names, then you will need to change the configuration to match your own ssh key pair filenames.

When you run the playbook against a cloud such as OpenStack or AWS, the ssh public key you generated above will be automatically injected into the servers during the Provision the servers step. If your servers were not provisioned by the Cello Ansible agent, you will have to manually inject/copy the public key onto each machine. For each machine, you will also need to place the ssh public key in the file named ~/.ssh/authorized_keys in the user account which you will use to log into the server.

Once you have the ssh keys, start up the ssh agent on the Ansible controller, using the following command:

    eval $(ssh-agent -s) && ssh-add ~/.ssh/fd

Clone Ansible agent code

The Ansible agent is part of the Hyperledger Cello project. To run its playbooks, you will need to clone the Cello repo.

   cd ~
   git clone --depth=1 https://gerrit.hyperledger.org/r/cello

This command will clone the repo into a folder named cello. The Ansible agent is at the following directory:

   ~/cello/src/agent/ansible

from this point on, you should stay in that directory for all command executions. This directory is also referred to root agent directory. For convenience, you can execute the following command to help you get back to this directory easily:

   export AAROOT=~/cello/src/agent/ansible
   cd $AAROOT

Install Ansible cloud module

When you run the playbook against a cloud such as OpenStack or AWS, Ansible requires that specific libraries be installed on the controller to access these clouds. These libraries are normally called Ansible cloud modules. When you work with a specific cloud, you will need only to install the cloud modules for that cloud. Here are the steps to install Ansible modules for AWS, Azure and OpenStack respectively:

    AWS:
    sudo pip install boto boto3

    Azure:
    sudo pip install azure

    OpenStack:
    sudo pip install shade

These modules are used during the Provision the servers step. If you are not running the Ansible agent against a cloud provider, you do not need any of these modules.

Create configuration files

The Ansible agent relies on two configuration files to work and stand up your fabric network. One is the cloud configuration file, the other is the fabric configuration file.

The cloud configuration file is used when running against a cloud such as AWS, Azure or OpenStack to create virtual machines, create security rules for networking, inject ssh keys, and eventually delete the virtual machines when you decide to destroy everything you created. Ansible agent provides the following sample cloud configuration files:

    $AAROOT/vars/aws.yml
    $AAROOT/vars/azure.yml
    $AAROOT/vars/os.yml

You should create a copy of one of the above cloud configuration example files and modify it as needed with your cloud credentials, etc.

The fabric configuration file is used to define how the Hyperledger Fabric network will be created. This file will let you specify what release of Fabric you want to use, what organizations to create, how many peers or orderers will be in an organization, how many kafka and zookeeper nodes you'd like to have and how you want to lay out your Fabric network on the servers you have. Ansible agent provides the following sample fabric configuration files:

    $AAROOT/vars/bc1st.yml
    $AAROOT/vars/bc2nd.yml
    $AAROOT/vars/vb1st.yml

You should create a copy of one of the above fabric configuration example files and modify it as needed with your desired Fabric topology, if needed.

Together, the fabric and cloud configuration files ultimately control what your Fabric network will look like. You should make your own copies, then make changes to those copies based on your own needs and credentials to run the Ansible playbook to produce your Fabric network. Your copies of these configuration files should be in the same directory as the example files, or Ansible will be unable to find them.

The following examples will assume you have created your own cloud configuration file copy named mycloud.yml and your own fabric configuration file copy named myfabric.yml from their respective sample files. Since both types of sample files are relatively large, have a good understanding of what each field means in both files is absolutely critical. Please refer to the following two sections for details on each field in the two types of files.

Cloud configuration file details
Fabric configuration file details

Provision the servers

This initial step provisions a set of virtual servers from cloud in a cloud provider.

    ansible-playbook -e "mode=apply env=mycloud cloud_type=os" provcluster.yml

The above command will provision (prov is short for provision) a cluster of virtual machines using an OpenStack cloud, with the environment and configuration defined in the vars/mycloud.yml file. The value apply for the parameter mode tells the playbook to create the resources. The value os for the parameter cloud_type indicates that we are running this playbook against an OpenStack cloud. The value mycloud for the parameter env indicates the cloud config file vars/mycloud.yml should be used. The possible values for mode are apply and destroy, the possible values for cloud_type are os, aws and azure at present.

This step produces a set of servers in your cloud and an Ansible host file named runhosts in this directory on your Ansible controller:

    $AAROOT/run

If you are working with servers already exist, you will need to follow the section Use the existing servers to continue setting up your fabric network.

To remove everything this step created, run the following command:

    ansible-playbook -e "mode=destroy env=mycloud cloud_type=os" provcluster.yml

Initialize the servers

This step will install all necessary software packages, setup an overlay network, and configure DNS services and registrator services on the machines created in previous step:

    ansible-playbook -i vars/runhosts -e "mode=apply env=mycloud env_type=flanneld" initcluster.yml

The parameter env is same as in previous step. The parameter env_type indicates what communication environment you would like to setup. The possible values for this parameter are flanneld and k8s. Value flanneld is used to setup a docker swarm like environment. Value k8s is to set up a Kubernetes environment.

To remove everything this step created, run the following command:

    ansible-playbook -i vars/runhosts -e "mode=destroy env=mycloud env_type=flanneld" initcluster.yml

Set up the Fabric network

This step will build (or download from a Docker repository) the various required Fabric binaries and docker images, create certificates, and eventually run various fabric components such as peer, orderer, kafka, zookeeper, fabric ca, etc on the environment produced by the previous steps:

    ansible-playbook -i vars/runhosts -e "mode=apply env=myfabric deploy_type=compose" setupfabric.yml

The env value in the command indicates which Fabric network configuration to use. The meaning of this parameter is a bit different compared to the previous commands. The parameter deploy_type determines if docker compose will be used to deploy, or Kubernetes will be used to deploy. This should corrlate to the env_type parameter given in the Initialize the servers step.

To remove everything this step created, run the following command:

    ansible-playbook -i vars/runhosts -e "mode=destroy env=myfabric deploy_type=compose" setupfabric.yml

Verify the Fabric network

If all previous steps run without any errors, you can run the following playbook to verify the running status of each container:

    ansible-playbook -i vars/runhosts -e "mode=verify env=bc1st" verify.yml

The env value in the command should match the value used in Setup the fabric network The command should access all the servers and display the container status for each container in your fabric network. If these containers do not exit/crash, then you know you have successfully deployed your own fabric network.

Useful tips for running the Ansible agent

Cloud configuration file details

The cloud configuration file is used by Ansible agent to work with a specific cloud. It is very important to make every field in the file accurate according to your own cloud. Most of the information in this file should be provided by your cloud provider. If you are not 100% sure what value a field should have, it would be a good idea to use the corresponding value in the sample cloud configuration file. The following section describes what each field means. Please use vars/os.yml as a reference and see example values for these fields:

auth: Authorization fields for a cloud
auth_url: url for cloud Authorization
username: User name to log in to the cloud account
password: Password for the user of the cloud account
project_name: Name of the project of your cloud, specific to OpenStack
domain: The domain of the project, specific to OpenStack
cluster: This section defines how virtual machines should be created
target_os: The operating system we are targeting, it has to be `ubuntu`
    at present
image_name: Cloud image name to be used to create virtual machines
region_name: Region name for VMs to reside in, leave blank if unsure
ssh_user: The user id to be used to access the virtual machines via ssh
availability_zone: The availability zone, leave blank to use default
validate_certs: When access cloud, should the certs to be validated?
    Set to false if your cloud use self signed certificate
private_net_name: The private network name from your cloud account on
    which your VMs should be created
flavor_name: Flavor name to create your virtual machine
public_key_file: The public ssh key file used to connect to servers,
    use absolute path
private_key_file: The private ssh key file, use absolute path,
    public_key_file and this key file should make a pair
node_ip: Use public ip or private ip to access each server, only possible
    value are `public_ip` and `private_ip`
assign_public_ip: Should each VM be allocated public IP address or not, true
    or false, default is true
container_network: This section defines overlay network, you should not
    change this unless you absolutely know what you are doing
Network: Overlay network address space, should be always in cidr notion,
    such as 172.16.0.0/16
SubnetLen: The bit length for subnets, it should be 24 normally, do not
    change it unless you absolutely know what you are doing
SubnetMin: minimum subnet
SubnetMax: maximum subnet
Backend: backend for flanneld setup
Type: the type for flanneld setup
Port: the port to use
service_ip_range: when use k8s, this defines service ip range
dns_service_ip: dns service ip address for k8s
node_ips: a list of public IP addresses if you like the VMs to be accessible
    and using preallocated IP addresses
name_prefix: VM name prefix when create new VMs, this combines with
    stack_size to make VM names. These names will be used in fabric
    configuration For example,if your prefix is fabric, and stack
    size is 3, then you will have 3 VMs named fabric001, fabric002,
    fabric003, these names will be referred as server logic names
domain: domain name to use when create fabric network nodes.
stack_size: how many VMs to be created
etcdnodes: which nodes you like the etcd to be set up on. only needed
    for k8s, should be a list of logic name like fabric001, fabric002,
    fabric003
builders: which VM to be used to build fabric binaries. should be only one
    machine, use logic name like fabric001, fabric002, etc.
flannel_repo: The url point to the flanneld tar gz file
etcd_repo: The url point to the etcd tar gz file
k8s_repo: The url point to the k8s binary root directory
go_repo: The url point to the go lang tar gz file
volume_size: when create VMs the size of the volume
block_device_name: block device name when create volume on OpenStack cloud
    fabric network, to verify that, you can run the following command to see

Fabric configuration file details

The Fabric configuration file defines how your Fabric network will look, how many servers you will use, how many organizations you would like to create, how many peers and orderers each organization has, and how many kafka and zookeeper containers will be set up. Additionally, it defines what names will be given to organizations, peers, orderers etc. This file defines the topology of your Fabric network, and a good understanding of this file is essential in order to create the Fabric network you need. Please use vars/bc1st.yml as a reference and see example values for these fields:

GIT_URL: hyperledger fabric git project url. should be always
    "http://gerrit.hyperledger.org/r/fabric"
GERRIT_REFSPEC: ref spec when build a specifc patch set. for example, it
    can be "refs/tags/v1.0.5"
fabric: This section define hyperledger fabric network layout
ssh_user: The user name to be used to log in to the remote servers
peer_db: The peer database type, possible values are CouchDB and leveldb
tls: Should this deployment use tls, default is false,
network: This section defines the layout of the fabric network
fabric001: This defines fabric containers running on the node named
    fabric001, each virtual or physical machine should have a section
    like this.
cas: list of the fabric certificate authority for an organization,
    the name of each ca should be in the format of <name>.<orgname>,
    for example, ["ca1st.orga", "ca1st.orgb"]
peers: list of the peers run on this node, the format of the names
    shuold be <role>@<name>.<orgname>, for example,
    ["anchor@peer1st.orga","worker@peer2nd.orga"], this means that
    there will be two peers running on this node, they are both from
    organization named orga, one is acting as
    anchor node, the other is the worker node.
orderers: list of the orderers run on this node, the format of the
    names should be <name>.<orgname>, for example, ["orderer1st.orgc",
    "orderer2nd.orgc"], this means that there will be two orderers
    running on this node, they are both from organization named orc,
    one is named orderer1st, and the other named orderer2nd.
zookeepers: list of the zookeeper containers run on this node. The
    format for zookeeper containers are <name>, since zookeeper
    containers do not belong to any organization, their names should
    be simply a string. For example: ["zookeeper1st", "zookeeper2nd"],
    this means that there will be two zookeeper containers running on
    this node, their names are zookeeper1st and zookeeper2nd respectively.
kafkas: list of the kafka containers run on this node The format for
    kafka containers are <name>, since kafka containers do not belong
    to any organization, their name should be simply a string. For
    example, ["kafka1", "kafka2"], this means that there will be two
    kafka containers running on this node, their names are kafka1 and
    kafka2.
baseimage_tag: docker image tag for fabric-peer, fabric-orderer,
    fabric-ccenv,fabric-tools. for example, it can be "x86_64-1.1.0-alpha",
    The value of this field is very important, if this value is empty,
    that means you like to build the fabric binaries and possibly docker
    container images. This field and the repo section determins where to
    download binaries or should binaries be downloaded.
helper_tag: docker image tag for container fabric-ca, fabric-kafka,
    fabric-zookeeper, for example, it be "x86_64-1.1.0-preview"
ca: This section defines how the fabric-ca admin user id and password
admin: ca user admin name
adminpw: ca admin user password
repo: This section defines where to get the fabric docker image and
    binary tar gz file. This allows you to use a local docker repository
url: Docker image repository for fabric, for example if you are using
    docker hub, the value will be "hyperledger/", if you are using
    nexus3, the value will be "nexus3.hyperledger.org:10001/hyperledger/"
bin: The url point to the fabric binary tar gz file which contains
    configtxgen, configtxlator, cryptogen etc.

K8S admin dashboard

Starting with Cello 0.8.0, Ansible agent has been upgraded to securely enable a k8s dashboard when you choose to deploy fabric network over k8s. The Ansible agent comes with a set of self-signed certificates in a directory named secrets/certs, if you do not want to use the default certificates, you should replace these certificates with your own. The Agent also creates a pair of users named admin and fabric. These are defined in the secrets/users/token.csv file. You can change and set your own passwords for these users here. Once you have everything set up, you should be able to access the k8s dashboard at the following url:

    https://<node_ip>>:32334/

When you are asked for the token, you can use either admintoken or fabrictoken to login.

Running an Ansible playbook

Ansible allows you to run tasks in a playbook with particular tags or skip particular tags. For example, you can run the follow command

    ansible-playbook -i vars/runhosts -e "mode=apply env=bc1st \
    deploy_type=compose" setupfabric.yml --tags "certsetup"

The above command will use the runhosts inventory file and only run tasks or plays tagged certsetup, all other plays in the playbooks will be skipped.

    ansible-playbook -i vars/runhosts -e "mode=apply env=bc1st \
    deploy_type=compose" setupfabric.yml --skip-tags "certsetup"

The above command will run all tasks except for the tasks/plays tagged certsetup

Setting up ssh-agent

Since Ansible's only means of communicating with the servers it configures is ssh, setting up ssh-agent on the Ansible controller is very important. If you do not do this, Ansible will likely fail to connect to your machines. Follow the steps below to set up ssh-agent on the Ansible controller (the machine you run the Ansible script on).

  1. Create a ssh key pair (only do this once):

    ssh-keygen -t rsa -f ~/.ssh/fd
    
  2. Run this command once in the active shell session you will run the Ansible script from :

    eval $(ssh-agent -s)
    ssh-add ~/.ssh/fd
    
  3. For the servers created via cloud providers, this step is already done for you. For existing servers, you will need to make sure that the fd.pub key is in the file ~/.ssh/authorized_keys. Otherwise, the servers will reject the ssh connection from Ansible controller.

Convenient configurations and commands

At the root directory of the Ansible agent there are set of preconfigured playbooks, they were developed as a convenient sample playbooks for you if you mainly work with a particular cloud. Here's a list of these playbooks.

aws.yml
awsk8s.yml
os.yml
osk8s.yml
vb.yml
vbk8s.yml

These files use their corresponding cloud and fabric configuration files. For example, aws.yml uses vars/aws.yml and vars/bc1st.yml to set up a multi-node Fabric network on AWS. awsk8s.yml uses vars/aws.yml and vars/bc1st.yml to set up a multi-node Fabric network on AWS within a k8s cluster. To use these playbooks, you simply need to make small changes in the coresponding configuration files in the vars directory, then issue the following command:

To stand up a Fabric network on AWS:

    ansible-playbook -e "mode=apply" aws.yml

To destroy a Fabric network on AWS:

    ansible-playbook -e "mode=destroy" aws.yml

If your target environment is OpenStack, then you will be using a slightly different command:

    ansible-playbook -e "mode=apply" os.yml or osk8s.yml
    ansible-playbook -e "mode=destroy" os.yml or osk8s.yml

Using existing servers

When you have a set of physical servers or virtual machines already available, you can still use the Ansible agent to stand up your Fabric network. To do that, you will need to manually configure some provisioning steps.

There are two things you need to do, one is to ensure that your servers can be accessed via ssh, the second is to produce a runhosts file like the one below. The hostnames of these servers have to form a patten using a prefix with three digits, for example, fabric001, fabric002, fabric003. The word fabric serves as a default prefix which can be changed to any string in the cloud configuration file. After the prefix, the three digit postfix should start at 001, and increment up to the defined stack size. In the below example, the prefix is fabric, but you can use any string you prefer as long as it is the same as the cloud configuration file's name_prefix field:

    cloud ansible_host=127.0.0.1
    169.45.102.186 private_ip=10.0.10.246 public_ip=169.45.102.186 inter_name=fabric001
    169.45.102.187 private_ip=10.0.10.247 public_ip=169.45.102.187 inter_name=fabric002
    169.45.102.188 private_ip=10.0.10.248 public_ip=169.45.102.188 inter_name=fabric003

    [allnodes]
    169.45.102.186
    169.45.102.187
    169.45.102.188

    [etcdnodes]
    169.45.102.186
    169.45.102.187
    169.45.102.188

    [builders]
    169.45.102.186

The above file is a typical Ansible host file. The cloud ansible_host should be your Ansible controller server, you should not change that line. All other lines in the file represent a server, private_ip and public_ip are a concept that only applies to cloud providers. If your servers are not in a cloud, then you can use the server's IP address for both private_ip and public_ip field, but you cannot remove these two fields. The inter_name field is also important, you should name the server sequentially and these names will be used in later configuration to allocate Hyperledger Fabric components. Group allnodes should list all the servers other than the Ansible controller node. Group etcdnodes should list all the servers that you wish to install etcd services on. Group builders should contain just one server that you wish to use to build Hyperledger Fabric artifacts such as executables and docker images.

Using existing k8s cluster

Ansible agent allows you to deploy fabric network onto an existing k8s cluster. To do that, you only need to place your k8s configuration file and possibly certificates in the ansible agent vars directory, the kube configuration file must be named kubeconfig. Then you can use exactly same fabric network configuration file such as bc1st.yml, vb1st.yml file to stand up your own fabric network. You can use the sample configuration file like bc1st.yml or you can use these sample configuration as a starting point to create new ones. Then you simply run the following command to stand up your own fabric network.

    ansible-playbook -e "mode=apply env=bc1st deploy_type=k8s" setupfabric.yml

The above command will use the vars/kubeconfig file to deploy fabric network defined in vars/bc1st.yml file.

To destroy what you just created, run the following command::

    ansible-playbook -e "mode=destroy env=bc1st deploy_type=k8s" setupfabric.yml

You can also choose to use the ansible agent container to run the above command.

Required Ports And Security Considerations

When you work with the public cloud, it is important to open or close certain ports for security and communication reasons. The following ports are used by the flanneld overlay network and other services of the Fabric network. You will need to make sure that the ports are open. The following example assumes that the overlay network is 10.17.0.0/16 and the docker host network is 172.31.16.0/20:

Custom UDP Rule  UDP  8285              10.17.0.0/16
Custom UDP Rule  UDP  8285              172.31.16.0/20
SSH              TCP  22                0.0.0.0/0
Custom TCP Rule  TCP  2000 - 60000      10.17.0.0/16
Custom TCP Rule  TCP  2000 - 60000      172.31.16.0/20
DNS (UDP)        UDP  53                172.31.16.0/20
DNS (UDP)        UDP  53                10.17.0.0/16
All ICMP - IPv4  All  N/A               0.0.0.0/0

Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 International License.