It is a very common task to pull your application code to EC2 instances from the Git repository. If you’re using CodeCommit as your main Git repository and CloudFormation for infrastructure management, it is very easy to launch an instance and allow it to access to that repository without storing any credentials or keys inside of it. In this article I’ll show you how to implement this in real life.
The Task
Our task is to automate initial data import from CodeCommit repository to dev MongoDB server during CloudFormation stack deployment.
Solution
To solve this task we need to do several things:
- Create MongoDB AMI
- Update your existing CloudFormation template
Easy.
MongoDB AMI
Let’s build our MongoDB AMI on top of Ubuntu 16.04 using Packer in N. Virginia Region. Here’s the template (mongodb.json):
"variables":
"aws_profile": "default"
,
"builders": [
"type": "amazon-ebs",
"profile": " user `aws_profile`",
"ami_name": "mongodb_server-timestamp",
"instance_type": "t2.small",
"source_ami": "ami-66506c1c",
"ssh_username": "ubuntu"
],
"provisioners": [
sudo tee /etc/apt/sources.list.d/mongodb-org-3.2.list",
"sudo apt-get update",
"sudo apt-get install -y mongodb-org python-pip",
"sudo systemctl enable mongod",
"sudo pip install awscli boto3"
]
,
"type": "file",
"source": "./configs/mongod.conf",
"destination": "/tmp/"
,
"type": "shell",
"inline":[
"sudo mv /tmp/mongod.conf /etc/"
]
]I’m also installing aws cli inside to be able to automate the things which are not covered by this article.
To build this template use the following command:
$ packer build mongodb.jsonCloudFormation template
First all you need to add an Instance declaration:
"MongoDB":
"Type": "AWS::EC2::Instance",
"Metadata" :
"Comment1" : "Restores MongoDB backup from CodeCommit repository",
"AWS::CloudFormation::Init" :
"configSets" :
"InstallCFN": [ "config-cfn-hup" ]
,
"config-cfn-hup":
"/etc/cfn/cfn-hup.conf" :
"content":
"Fn::Join": [
"",
[
"[main]\n",
"stack=", "Ref" : "AWS::StackId" , "\n",
"region=", "Ref" : "AWS::Region" , "\n"
]
]
,
"mode" : "000400",
"owner" : "root",
"group" : "root"
,
"/etc/cfn/hooks.d/cfn-auto-reloader.conf":
"content":
"Fn::Join" : [
"", [
"[cfn-auto-reloader-hook]\n",
"triggers=post.update\n",
"path=Resources.MongoDB.Metadata.AWS::CloudFormation::Init\n",
"action=/opt/aws/bin/cfn-init -v ",
" --stack ", "Ref" : "AWS::StackName" ,
" --resource MongoDB ",
" --configsets InstallCFN ",
" --region ", "Ref" : "AWS::Region" , "\n",
"runas=root\n"
]
]
,
"/etc/systemd/system/cfn-hup.service":
"content":
"Fn::Join" : [
"", [
"[Unit]\n",
"Description=Cloud formation helper daemon\n",
"\n",
"[Service]\n",
"ExecStart=/usr/local/bin/cfn-hup\n",
"Restart=always\n",
"RestartSec=10s\n",
"Type=notify\n",
"NotifyAccess=all\n",
"TimeoutStartSec=120\n",
"TimeoutStopSec=15\n",
"\n",
"[Install]\n",
"WantedBy=multi-user.target\n"
]
]
,
"commands" :
"enable-cfn-hup" :
"command": "systemctl enable cfn-hup.service"
,
"start-cfn-hup":
"command": "systemctl start cfn-hup.service"
,
"Properties":
"SubnetId":
"Ref": "PublicSubnet0"
,
"ImageId":
"Ref": "MongoDBImage"
,
"InstanceType": "t2.small",
"KeyName":
"Ref": "SSHKeyName"
,
"SecurityGroupIds": [
"Ref": "SecurityGroupMongoDB"
],
"IamInstanceProfile":
"Ref": "MongoDBServerInstanceProfile"
,
"UserData":
"Fn::Base64": sed 's/.json$//'
,
"Tags": [
"Key": "Name",
"Value": "MongoDB"
,
"Key": "Application",
"Value":
"Ref": "AWS::StackId"
]Such MongoDB instance declaration will install cfn-init, and put some additional automation logic to user-data. The most important of instance parameters is IamInstanceProfile, which will apply appropriate IAM Role to our server (we will add it later).
We’re passing all automation logic through instance user-data. It consists of 3 parts:
- Install cfn-init
- Clone repository with MongoDB backups and restore them only once during instance first boot.
- Launch cfn-init automation, if you’d like to add something.
I moved out MongoDB backup import to user-data because cfn-init did not allow me to launch and configure git properly.
All git the magic happens here:
# Using AWS cli to get temporary credentials to CodeCommit
$ git config --global credential.helper '!aws codecommit credential-helper $@'
# Configuring git to use HTTP proto
$ git config --global credential.UseHttpPath true
# Cloning our repo
$ git clone https://git-codecommit.us-east-1.amazonaws.com/v1/repos/mongodb-backupIAM instance profile
In order to allow everything to work properly you need to create IAM instance profile with the right permissions. Here it is:
"MongoDBServerRole":
"Type": "AWS::IAM::Role",
"Properties":
"AssumeRolePolicyDocument":
"Version" : "2012-10-17",
"Statement": [
"Effect": "Allow",
"Principal":
"Service": [
"ec2.amazonaws.com"
]
,
"Action": [
"sts:AssumeRole"
]
]
,
"Path": "/",
"RoleName":
"Fn::Join": [
"-", [
"MongoDBServerRole",
"Ref": "AWS::StackName"
]
]
,
"MongoDBServerPolicy":
"Type" : "AWS::IAM::Policy",
"Properties" :
"PolicyName" :
"Fn::Join": [
"-", [
"MongoDBServerPolicy",
"Ref": "AWS::StackName"
]
]
,
"PolicyDocument" :
"Version": "2012-10-17",
"Statement": [
"Effect": "Allow",
"Action": "codecommit:ListRepositories",
"Resource": "*"
,
"Effect": "Allow",
"Action": "codecommit:*",
"Resource": "arn:aws:codecommit:us-east-1:468439730987:mongodb-backup"
,
"Effect": "Allow",
"Action": [
"kms:Encrypt",
"kms:Decrypt",
"kms:ReEncrypt",
"kms:GenerateDataKey",
"kms:GenerateDataKeyWithoutPlaintext",
"kms:DescribeKey"
],
"Resource": "*"
]
,
"Roles": [
"Ref": "MongoDBServerRole"
]
,
"MongoDBServerInstanceProfile":
"Type": "AWS::IAM::InstanceProfile",
"Properties":
"Path": "/",
"Roles": [
"Ref": "MongoDBServerRole"
],
"InstanceProfileName":
"Fn::Join": [
"-", [
"MongoDBServerInstanceProfile",
"Ref": "AWS::StackName"
]
]Final words
In this article we saw how easily we could build our own AMI using Packer and organize it’s access to CodeCommit repository without storing any credentials inside the instance or it’s environment. IAM roles and instance configurations are provided as CloudFormation template. Hope, this will help you to save some time for a cup of coffee.
