Skip to content

Latest commit

 

History

History
230 lines (166 loc) · 10.4 KB

README.md

File metadata and controls

230 lines (166 loc) · 10.4 KB

Table of Contents

Custom Images for Cloudbreak

What is Cloudbreak?

Cloudbreak is a tool to simplify the provisioning, configuration and scaling of Hortonworks Data Platform clusters on cloud provider infrastructure. Cloudbreak can be used to provision across cloud infrastructure providers including: Amazon Web Services (AWS), Microsoft Azure and Google Cloud Platform (GCP).

Learn more about Cloudbreak here: http://hortonworks.github.io/cloudbreak-docs/

What are Custom Images?

Cloudbreak launches clusters from an image that includes default configuration and default tooling for provisioning. These are considered the Standard Default images and these images are provided with each Cloudbreak version.

From bird's-eye view, images contain the following:

  • Operating system (e.g. CentOS, Amazon Linux)
  • Standard configuration (disabled SE Linux, permissive iptables, best practice configs, etc.)
  • Standard tooling (bootstrap scripts, bootstrap binaries)

Important: Ambari and HDP packages are not part of the image and the desired version of Ambari and HDP packages are downloaded during provision time. This makes the images agnostic to the version of Ambari and HDP that can be installed by Cloudbreak.

The following Standard Default images and Linux versions are available for each Cloudbreak version:

  • Amazon: CentOS 7.6
  • Azure: CentOS 7.6

In some cases, these default images might not fit the requirements of users (e.g. they need custom OS hardening, libraries, tooling, etc) and instead, the user would like to start their clusters from their own custom image. The repository includes instructions and scripts to help build those custom images. Once you have an images, refer to the Cloudbreak documentation for information on how to register and use these images with Cloudbreak: http://hortonworks.github.io/cloudbreak-docs/

You have two options for building a custom image:

  • If you are only looking to change the OS and use a Cloudbreak base image for that OS, use the instructions below for your cloud provider and build the image for a different OS, or
  • If you want to start from your own base image, follow the instructions in Advanced topics to modify the package.json to start from your own base image.

Using this Repository

Our recommendation is to fork this repo to to your own GitHub account or to the account of your organization and you can make changes there and create an image from there. If you think that some of the changes you made might be useful for the Cloudbreak product as a whole, feel free to send us a pull request.

Note: After you have have forked the repository, you are responsible to keep it up to date and fetch the latest changes from the upstream repository.

Finding the Correct Branch

This repository contains different branches for different Cloudbreak versions. Cloudbreak versions are defined as:

<major>.<minor>.<patch>[-build sequence] e.g 1.16.3 or 1.16.4-rc.7 or 2.23.0-b153

If you are creating a custom image for Cloudbreak, always make sure that you are using the correct branch from cloudbreak-images repository. You can find the related branch based on the and version numbers of Cloudbreak (e.g if you are using 1.16.3 or 1.16.4-rc.7 version of Cloudbreak then the related branch is rc-1.16). If you are using 2.0.1 version of Cloudbreak then the related image branch is rc-2.0. If you are using 2.23.0-b153 version of Cloudbreak then the related branch is CB-2.23.0

Note: If you do not use the appropriate branch for creating your image then there is a chance that Cloudbreak will not be able to install the cluster successfully.

Building a Custom Image

Packer

Images for Cloudbreak are created by Packer. The main entry point for creating an image is the Makefile which provides wrapper functionality around Packer scripts. You can find more details about how it works in the Packer documentation.

Main configuration of Packer for building the Cloudbreak images is located in the packer.json file.

Prerequisites

The following are requirements for the image building environment:

AWS

Set the following environment variables to build AWS images:

  • AWS_ACCESS_KEY_ID
  • AWS_SECRET_ACCESS_KEY

Example for environment variables:

export AWS_ACCESS_KEY_ID=AKIAIQ**********
export AWS_SECRET_ACCESS_KEY=XHj6bjmal***********************

Note: Since Packer is the underlaying technology used to build the AWS images, you can learn more about the environment variables at Packer > Amazon Reference and the minimal set of AWS IAM or Role policies necessary at Packer > Amazon EC2 Roles. A traditional IAM Role can be used in place of AWS Access Keys using the same policy definition.

Use the following commands to build AWS images based on the following base operating systems:

OS Build Command
CentOS 7 make build-aws-centos7

If you want to start from your own base image, follow the instructions in Advanced topics to modify the package.json to start from your own base image. Then use the commands above to build that image.

Permissions

The created Snapshots and AMIs will be private on AWS by default, you could use the following environment variables to change this behavior:

#Use this to make the created AMIs public
export MAKE_PUBLIC_AMIS=yes

#Use this to make the created Snapshot of the AMIs public
export MAKE_PUBLIC_SNAPSHOTS=yes

Note: Environment variables need to be exported before the use of the make command.

Deployment Regions

The generated AMI will be deployed to multiple AWS Regions as listed in the Makefile. So be sure to review and edit the Makefile with updates to the AWS_AMI_REGIONS and AWS_GOV_AMI_REGIONS to limit deployment.

Azure

Set the following environment variables to build Azure images:

  • ARM_CLIENT_ID
  • ARM_CLIENT_SECRET
  • ARM_SUBSCRIPTION_ID
  • ARM_TENANT_ID
  • ARM_GROUP_NAME
  • ARM_STORAGE_ACCOUNT

Example for environment variables:

export ARM_CLIENT_ID=3234bb21-e6d0-*****-****-**********
export ARM_CLIENT_SECRET=2c8bzH******************************
export ARM_SUBSCRIPTION_ID=a9d4456e-349f-*****-****-**********
export ARM_TENANT_ID=b60c9401-2154-*****-****-**********
export ARM_GROUP_NAME=resourcegroupname
export ARM_STORAGE_ACCOUNT=storageaccountname

Note: Since Packer is the underlaying technology used to build the Azure images, you can learn more about the environment variables at Packer > Azure Reference.

Use the following commands to build Azure images based on the following base operating systems:

OS Build Command
CentOS 7 make build-azure-centos7

If you want to start from your own base image, follow the instructions in Advanced topics to modify the package.json to start from your own base image. Then use the commands above to build that image.

GCP

Set the following environment variables to build GCP images:

  • GCP_STORAGE_BUNDLE
  • GCP_PROJECT
  • GCP_ACCOUNT_FILE

Example for environment variables:

export GCP_STORAGE_BUNDLE==*****
export GCP_PROJECT=*****
export GCP_ACCOUNT_FILE={path for the GCP credential json}

Note: Since Packer is the underlaying technology used to build the GCP images, you can learn more about the environment variables at Packer > GCP Reference.

Use the following commands to build GCP images based on the following base operating systems:

OS Build Command
CentOS 7 make build-gc-centos7

If you want to start from your own base image, follow the instructions in Advanced topics to modify the package.json to start from your own base image. Then use the commands above to build that image.

Running packer in debug mode

If you run Packer in debug mode then you can SSH into the VM during build phase and do additional debugging steps on the VM. This is how to start a build in debug mode:

PACKER_OPTS=--debug make build-aws-centos7

In debug mode, you need to hit enter before each step is executed by Packer. Once the VM is launched by Packer, it places a temporary SSH key in the launch directory with which you can login and do additional debug steps:

ssh -i ec2_aws-rhel7.pem ec2-user@<address of the machine displayed by Packer>

Check the logs without debug mode

A simple file browser is launched during image creation which can be accessed on port 9999.

User: admin, password: secret.

To access the browser, you need to open port 9999 in the security group of the generated resource group manually on your cloud provider. The generated resource group name will be displayed at the start of the build process.

E.g. on Azure:

    arm-centos7: Creating Azure Resource Manager (ARM) client ...
==> arm-centos7: Creating resource group ...
==> arm-centos7:  -> ResourceGroupName : 'packer-Resource-Group-qx0lx7wkg7'
==> arm-centos7:  -> Location          : 'northeurope'
==> arm-centos7:  -> Tags              :

Advanced topics

You can read more about postprocessors and customizing your base image with custom scripts and logic here.

Image content

You can find more information about the packages of specified images to be installed here. If new package needs to be added to the different images or deleted for any reason, the package in the file above must be changed in the appropriate section.