Code Monkey home page Code Monkey logo

ocp-manual-sts's Introduction

Running pods in OpenShift with AWS IAM Roles for service accounts aka IRSA

By: Alejandro Guadarrama Dominguez

You can associate an IAM role with an OpenShift service account. This OpenShift service account can then be used to run a pod providing AWS permissions to the containers. With this feature, pods on OpenShift can call AWS APIs.

Pod applications must sign their AWS API requests with AWS Security Token Service (AWS STS) as a web service that enables you to request temporary, limited-privilege credentials for AWS Identity and Access Management (IAM). This feature provides a strategy for managing credentials for your applications. The applications in the pod’s containers can then use an AWS SDK or the AWS CLI to make API requests to authorized AWS services.

The IAM roles for service accounts feature provides the following benefits:

  • Least privilege — By using the IAM roles for service accounts feature, you no longer need to provide extended permissions to the node IAM role so that pods on that node can call AWS APIs. You can scope IAM permissions to a service account, and only pods that use that service account have access to those permissions.
  • Credential isolation — A container can only retrieve credentials for the IAM role that is associated with the OpenShift service account and namespace to which it belongs. A container never has access to credentials that are intended for another container that belongs to another pod or namespace.
  • Auditability — Access and event logging is available through AWS CloudTrail to help ensure retrospective auditing.

To enable the use of IAM Roles for OpenShift service accounts, you should configure an AWS cluster in manual mode to use Amazon Web Services Secure Token Service (AWS STS). With this configuration, the CCO uses temporary credentials for different components.

alt_text

To deploy a cluster with CCO in manual mode with STS, you can check the documentation https://docs.openshift.com/container-platform/4.8/authentication/managing_cloud_provider_credentials/cco-mode-sts.html. This blog does not show how to do that. We will focus on how to use the feature to run pods with temporary, limited-privilege credentials provided by AWS STS.

ROSA Clusters

To install a ROSA cluster with STS follow the documentation

NOTE: For ROSA Clusters already configured with STS, using the Option 2

Prepare IAM roles and link with OpenShift service accounts

Option 1: using ccoctl tool

This is an example of a CredentialsRequest resource to create any role to use for an OpenShift service account.

apiVersion: cloudcredential.openshift.io/v1
kind: CredentialsRequest
metadata:
  labels:
    controller-tools.k8s.io: "1.0"
  name: manual-sts
  namespace: manual-sts
spec:
  providerSpec:
    apiVersion: cloudcredential.openshift.io/v1
    kind: AWSProviderSpec
    statementEntries:
    - action:
      - s3:Get*
      - s3:Put*
      effect: Allow
      resource: '*'
  secretRef:
    name: manual-sts
    namespace: manual-sts
  serviceAccountNames:
  - sa-manual-sts

It is very important to create the role with the ccoctl tool. This step is to create the IAM role and link internally to the OpenShift service account and namespace. Once it is created, the IAM role can be modified by adding or removing permissions in AWS IAM Console, running AWS IAM commands, or using an automation tool like Ansible or Terraform.

The ccoctl tool can create one or more CredentialsRequest. The tool receives a folder as parameter and processes all the CredentialsRequest yaml files in the folder. In this case, we save the file in the credrequests folder.

[root@bastion ~]# ccoctl aws create-iam-roles \
  --name=prefix-role \
  --region=us-east-1 \
  --credentials-requests-dir=credrequests  \
--identity-provider-arn=arn:aws:iam::1234567890:oidc-provider/manual-sts-oidc.s3.us-east-1.amazonaws.com  \
  --output-dir=outputs

The command above will create a file or files with an OpenShift secret being applied in the namespace. The identity provider is the value in the pre-installation tasks, or you can get in the amazon console IAM -> Identity providers.

outputs/manual-sts-manual-sts-credentials.yaml.

apiVersion: v1
stringData:
  credentials: |-
    [default]
    role_arn = arn:aws:iam::1234567890:role/prefix-role-manual-sts-manual-sts
    web_identity_token_file = /var/run/secrets/openshift/serviceaccount/token
kind: Secret
metadata:
  name: manual-sts
  namespace: manual-sts
type: Opaque

Also, we need to create the service(s) account(s) in the namespace.

[root@bastion ~]# oc create sa sa-manual-sts

How to assume the role

Below is an example of a POD running the aws-cli to test aws sts assume-role-with-web-identity.

The pod is placing the credentials in the file /root/.aws/ config and setting 2 environment variables AWS_ROLE_SESSION_NAME and AWS_REGION.

This example is not mounting directly the volume with the config file, because the path /root/.aws must be writable to let the aws-cli create the folder /root/.aws/cli/cache and store the temporary credentials if you are running aws-cli commands directly.

The pod has 3 requirements:

  1. Run with the service account in the CredentialsRequest
  2. Mount a volume with the secret generated after create the CredentialsRequest
  3. Mount the service account token with the audience openshift
apiVersion: v1
kind: Pod
metadata:
  annotations:
  labels:
    app: manual-sts
  name: manual-sts
  namespace: manual-sts
spec:
  containers:
  - image: amazon/aws-cli
    command: ["/bin/sh"]
    args: ["-c", "mkdir /root/.aws; cat /var/run/secrets/cloud/credentials > /root/.aws/config; sleep 86400"]
    imagePullPolicy: IfNotPresent
    name: manual-sts
    env:
    - name: AWS_ROLE_SESSION_NAME
      value: manual-sts-session
    - name: AWS_REGION
      value: us-east-1
    ports:
    - containerPort: 80
      name: http
      protocol: TCP
    resources:
      requests:
        cpu: 100m
        memory: 512Mi
    volumeMounts:
    - mountPath: /var/run/secrets/cloud
      name: manual-sts
      readOnly: false
    - mountPath: /var/run/secrets/openshift/serviceaccount
      name: bound-sa-token
      readOnly: true
  dnsPolicy: ClusterFirst
  serviceAccount: sa-manual-sts #<--The pod must run with the sa
  serviceAccountName: sa-manual-sts
  volumes:
  - name: manual-sts
    secret:
      defaultMode: 420
      optional: false
      secretName: manual-sts #<--This volume mount the config
  - name: bound-sa-token #<--The second volumen mount the web identity token
    projected:
      defaultMode: 420
      sources:
      - serviceAccountToken:
          audience: openshift
          expirationSeconds: 3600
          path: token

With the /root/.aws/config in the pod file, we can run any AWS command, The client will request the assume role in the background and will create the temporary credentials in the directory /root/.aws/cli/cache

[root@bastion ~]# ls -l /root/.aws/
total 4
-rw-r--r--. 1 root root 157 Jan 1 20:44 config

[root@bastion ~]# aws s3 ls
2020-11-13 18:01:43 fake1-buket
2021-10-03 18:01:39 fake2-buket
2021-11-19 01:13:29 fake3-buket
2022-12-18 18:02:45 manual-sts-oidc
2021-12-13 18:02:45 fake4-buket
[root@bastion ~]#

[root@bastion ~]# ls -l /root/.aws/
total 4
drwxr-xr-x. 3 root root  19 Jan 1 23:12 cli
-rw-r--r--. 1 root root 157 Jan 1 20:44 config

[root@bastion ~]# ls -l /root/.aws/cli/cache/
total 4
-rw-------. 1 root root 1887 Jan 1 23:12 234z3583d987653r6h3453455d5967b345529zs.json



[root@bastion ~]# cat /root/.aws/cli/cache/234z3583d987653r6h3453455d5967b345529zs.json | jq .
{
  "Credentials": {
    "AccessKeyId": "FAKEKEYID",
    "SecretAccessKey": "FAKESECRETBLABLABLA",
    "SessionToken": "FAKETOKEN",
    "Expiration": "2022-01-01T13:13:13+00:00"
  },
  "SubjectFromWebIdentityToken": "system:serviceaccount:manual-sts:sa-manual-sts",
  "AssumedRoleUser": {
    "AssumedRoleId": "IORETJOGFNDLFKGJIOWKD:manual-sts-session",
    "Arn": "arn:aws:sts::0987654321:assumed-role/prefix-role-manual-sts-manual-sts/manual-sts-session"
  },
  "Provider": "arn:aws:iam::0987654321:oidc-provider/manual-sts-oidc.s3.us-east-1.amazonaws.com",
  "Audience": "openshift",
  "ResponseMetadata": {
    "RequestId": "9fddf440-80b6-11ec-a8a3-0242ac120002",
    "HTTPStatusCode": 200,
    "HTTPHeaders": {
      "x-amzn-requestid": "9fddf440-80b6-11ec-a8a3-0242ac120002",
      "content-type": "text/xml",
      "content-length": "2076",
      "date": "Sat, 1 Jan 2022 07:13:13 GMT"
    },
    "RetryAttempts": 0
  }
}

What happened here? The aws-cli in the background tried 3 tasks:

  1. Try to find the ~/.aws./credentials file if is not possible
  2. Try to get the role from the metadata from http://169.254.169.254/latest/meta-data/ if is not possible
  3. At the end, based on the file in ~/.aws/config (see below), aws-cli try to assume the role provided in combination with the environment variables that we set in the pod; AWS_REGION and AWS_ROLE_SESSION_NAME
[root@bastion ~]# cat /root/.aws/config
[default]
role_arn = arn:aws:iam::0987654321:role/prefix-role-manual-sts-manual-sts

Another option is run the command aws iam assume-role-with-web-identity. It is the similar behavior when you use any language SDK, like boto3 in Python.

[root@bastion]# TOKEN=$(cat /var/run/secrets/openshift/serviceaccount/token)
[root@bastion]# ARN=arn:aws:iam::0987654321:role/prefix-role-manual-sts-manual-sts
[root@bastion]# aws sts assume-role-with-web-identity \
  --role-arn $ARN \
  --role-session-name $AWS_ROLE_SESSION_NAME  \
  --web-identity-token $TOKEN \
  --duration-seconds 900
{
    "Credentials": {
        "AccessKeyId": "FAKEKEYID",
        "SecretAccessKey": "FAKESECRETBLABLABLA",
        "SessionToken": "FAKETOKEN",
        "Expiration": "2022-01-01T13:13:13+00:00"
    },
    "SubjectFromWebIdentityToken": "system:serviceaccount:manual-sts:sa-manual-sts",
    "AssumedRoleUser": {
        "AssumedRoleId": "SDLKJSDFKJSDFKLSJFDSK:manual-sts-session",
        "Arn": "arn:aws:sts::0987654321:assumed-role/prefix-role-manual-sts-manual-sts/manual-sts-session"
    },
    "Provider": "arn:aws:iam::0987654321:oidc-provider/manual-sts-oidc.s3.us-east-1.amazonaws.com",
    "Audience": "openshift"
}

Option 2: using aws-pod-identity-webhook

The ccoctl tool, as you can see, helps to create the AWS IAM Roles one or multiples with only one line of code. In cases where another area is responsible to manage the AWS IAM Roles, we can split the process.

  • Other area: create AWS IAM Roles
  • OpenShift: create service accounts and pods

On the side of AWS, we create the role and modify the “trust policy” to allow the pod service account to use it and attach the policy role to grant the desired permissions to the role.

{
 "Version": "2012-10-17",
 "Statement": [
  {
   "Effect": "Allow",
   "Principal": {
    "Federated": "arn:aws:iam::0987654321:oidc-provider/manual-sts-oidc.s3.us-east-1.amazonaws.com"
   },
   "Action": "sts:AssumeRoleWithWebIdentity",
   "Condition": {
    "StringEquals": {
     "manual-sts-oidc.s3.us-east-1.amazonaws.com:sub": "system:serviceaccount:manual-sts:sa-manual-sts"
    }
   }
  }
 ]
}

Create the service account

The service account requires some annotations, as you can see below, which is necessary for the webhook to set some environment variables inside the pod in an automatic way.

apiVersion: v1
kind: ServiceAccount
metadata:
  name: sa-manual-sts
  namespace: manual-sts
  annotations:
    eks.amazonaws.com/role-arn: "arn:aws:iam::077896975944:role/poc-irsa-s3-s3-sample-s3-sample-1"
    # optional: Defaults to "sts.amazonaws.com" if not set
    eks.amazonaws.com/audience: "sts.amazonaws.com"
    # optional: When set to "true", adds AWS_STS_REGIONAL_ENDPOINTS env var
    #   to containers
    eks.amazonaws.com/sts-regional-endpoints: "true"
    # optional: Defaults to 86400 for expirationSeconds if not set
    #   Note: This value can be overwritten if specified in the pod
    #         annotation as shown in the next step.
    eks.amazonaws.com/token-expiration: "86400"
    # the minimum time is 15 minutes "900"

Creating the pod

apiVersion: v1
kind: Pod
metadata:
  annotations:
  labels:
    app: manual-sts
  name: manual-sts
  namespace: manual-sts
spec:
  containers:
  - image: amazon/aws-cli
    command: ["/bin/sh"]
    args: ["-c", "sleep 86400"]
    imagePullPolicy: IfNotPresent
    name: manual-sts
    env:
    - name: AWS_ROLE_SESSION_NAME
      value: manual-sts-session
    - name: AWS_REGION
      value: us-east-1
    ports:
    - containerPort: 80
      name: http
      protocol: TCP
    resources:
      requests:
        cpu: 100m
        memory: 512Mi
  dnsPolicy: ClusterFirst
  serviceAccount: sa-manual-sts #<--The pod must run with the sa
  serviceAccountName: sa-manual-sts

After deploying the pod, the annotations in the service account with the aws-pod-identity-webhook, will inject the following env: variables merging with the existing ones and the volume and volumeMounts necessary to AssumeRoleWithWebIdentity.

    env:
    - name: AWS_DEFAULT_REGION
      value: us-west-2
    - name: AWS_REGION
      value: us-west-2
    - name: AWS_ROLE_ARN
      value: "arn:aws:iam::111122223333:role/s3-reader"
    - name: AWS_WEB_IDENTITY_TOKEN_FILE
      value: "/var/run/secrets/eks.amazonaws.com/serviceaccount/token"
    - name: AWS_STS_REGIONAL_ENDPOINTS
      value: "regional"
    volumeMounts:
    - mountPath: "/var/run/secrets/eks.amazonaws.com/serviceaccount/"
      name: aws-token
  volumes:
  - name: aws-token
    projected:
      sources:
      - serviceAccountToken:
          audience: "sts.amazonaws.com"
          expirationSeconds: 86400
          path: token

As you can see from the two examples above, it is very easy and transparent to use the IAM roles to call AWS services from a pod using the CCO in manual mode with STS.

ocp-manual-sts's People

Contributors

alexgdmx avatar

Watchers

 avatar  avatar

Recommend Projects

  • React photo React

    A declarative, efficient, and flexible JavaScript library for building user interfaces.

  • Vue.js photo Vue.js

    🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.

  • Typescript photo Typescript

    TypeScript is a superset of JavaScript that compiles to clean JavaScript output.

  • TensorFlow photo TensorFlow

    An Open Source Machine Learning Framework for Everyone

  • Django photo Django

    The Web framework for perfectionists with deadlines.

  • D3 photo D3

    Bring data to life with SVG, Canvas and HTML. 📊📈🎉

Recommend Topics

  • javascript

    JavaScript (JS) is a lightweight interpreted programming language with first-class functions.

  • web

    Some thing interesting about web. New door for the world.

  • server

    A server is a program made to process requests and deliver data to clients.

  • Machine learning

    Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.

  • Game

    Some thing interesting about game, make everyone happy.

Recommend Org

  • Facebook photo Facebook

    We are working to build community through open source technology. NB: members must have two-factor auth.

  • Microsoft photo Microsoft

    Open source projects and samples from Microsoft.

  • Google photo Google

    Google ❤️ Open Source for everyone.

  • D3 photo D3

    Data-Driven Documents codes.