HashiCorp Vault

Build Status Jenkins Plugin GitHub release Jenkins Plugin Installs

This plugin adds a build wrapper to set environment variables from a HashiCorp Vault secret. Secrets are generally masked in the build log, so you can't accidentally print them.
It also has the ability to inject Vault credentials into a build pipeline or freestyle job for fine-grained vault interactions.

Vault Authentication Backends

This plugin allows authenticating against Vault using the AppRole authentication backend. Hashicorp recommends using AppRole for Servers / automated workflows (like Jenkins) and using Tokens (default mechanism, Github Token, ...) for every developer's machine. Furthermore, this plugin allows using a Github personal access token, or a Vault Token - either configured directly in Jenkins or read from an arbitrary file on the Jenkins controller.

How does AppRole work?

In short: you register an approle auth backend using a self-chosen name (e.g. Jenkins). This approle is identified by a role-id and secured with a secret_id. If you have both of those values you can ask Vault for a token that can be used to access vault. When registering the approle backend you can set a couple of different parameters:

  • How long should the secret_id live (can be indefinite)
  • how often can one use a token that is obtained via this backend
  • which IP addresses can obtain a token using role-id and secret-id?
  • many more

This is just a short introduction, please refer to Hashicorp itself to get detailed information.

Isolating policies for different jobs

It may be desirable to have jobs or folders with separate Vault policies allocated. This may be done with the optional policies configuration option combined with authentication such as the AppRole credential. The process is the following:

  • The Jenkins job attempts to retrieve a secret from Vault
  • The AppRole authentication is used to retrieve a new token (if the old one has not expired yet)
  • The Vault plugin then uses the policies configuration value with job info to come up with a list of policies
  • If this list is not empty, the AppRole token is used to retrieve a new token that only has the specified policies applied
  • This token is then used for all Vault plugin operations in the job

The policies list may be templatized with values that can come from each job in order to customize policies per job or folder. See the policies configuration help for more information on available tokens to use in the configuration. The Limit Token Policies option must also be enabled on the auth credential. Please note that the AppRole (or other authentication method) should have all policies configured as token_policies and not identity_policies, as job-specific tokens inherit all identity_policies automatically.

What about other backends?

Hashicorp explicitly recommends the AppRole Backend for machine-to-machine authentication. Token based auth is mainly supported for backward compatibility. Other backends that might make sense are the AWS EC2 backend, the Azure backend, and the Kubernetes backend. But we do not support these yet. Feel free to contribute!

Implementing additional authentication backends is actually quite easy:

Simply provide a class extending AbstractVaultTokenCredential that contains a Descriptor extending BaseStandardCredentialsDescriptor. The Descriptor needs to be annotated with @Extension. Your credential needs to know how to authenticate with Vault and provide an authenticated Vault session. See VaultAppRoleCredential.java for an example.

Plugin Usage

Configuration

You can configure the plugin on three different levels:

  • Global: in your global config
  • Folder-Level: on the folder your job is running in
  • Job-Level: either on your freestyle project job or directly in the Jenkinsfile

The lower the level the higher its priority, meaning: if you configure a URL in your global settings, but override it in your particular job, this URL will be used for communicating with Vault. In your configuration (may it be global, folder or job) you see the following screen: Global Configuration

The credential you provide determines what authentication backend will be used. Currently, there are five different Credential Types you can use:

Vault App Role Credential

App Role Credential

You enter your role-id and secret-id there. The description helps to find your credential later, the id is not mandatory (a UUID is generated by default), but it helps to set it if you want to use your credential inside the Jenkinsfile.

The path field is the approle authentication path. This is, by default, "approle" and this will also be used if no path is specified here.

Migrate current Jenkins Vault configuration to support a new version of plugin.

After update a plugin from version 2.2.0 you can note - builds failed with an exception java.lang.NullPointerException. These steps will help you fix it:

  1. If you using AppRole auth method - you need to update Jenkins Credential store (in UI) for all kinds Vault App Role Credential and set Path field for your correct path or just leave the default approle and save.
  2. Go to Configure of failed job and change Vault Engine in Advanced Settings and choose your version on KV Engine 1 or 2 from a select menu K/V Engine Version for ALL Vault Secrets and save.

Vault Github Credential

Github Credentia

You enter your github personal access token to authenticate to vault.

Vault GCP Credential

GCP Credential

You enter your Vault GCP auth role name and audience. The JWT will be automatically retrieved from GCE metdata. This requires that Jenkins master is running on a GCE instance.

Vault Kubernetes Credential

Kubernetes Credential

You enter your Vault Kubernetes auth role. The JWT will be automatically retrieved from the mounted secret volume (/var/run/secrets/kubernetes.io/serviceaccount/token). This assumes, that the jenkins is running in Kubernetes Pod with a Service Account attached.

Vault AWS IAM Credential

AWS IAM Credential

Authenticate to Vault using the aws auth method with the IAM workflow. The AWS credentials will be automatically retrieved from one of several standard locations. The typical use case would be Jenkins master running on an AWS EC2 instance with the credentials acquired from the instance metadata. Optionally enter your AWS IAM auth role name and Vault AWS auth mount path. If the role is not provided, Vault will determine it from the principal in the IAM identity. If the mount path is not provided, it defaults to aws.

Vault Token Credential

Token Credential

Directly specify a token to be used when authenticating with vault.

Vault Token File Credential

Token File Credential

Basically the same as the Vault Token Credential, just that the token is read from a file on your Jenkins Machine. You can use this in combination with a script that periodically refreshes your token.

Usage in FreeStyle Jobs

If you still use free style jobs (hint: you should consider migrating to Jenkinsfile), you can configure both configuration and the secrets you need on the job level.

Job Configuration

The secrets are available as environment variables then.

Usage via Jenkinsfile

Let the code speak for itself:

node {
    // define the secrets and the env variables
    // engine version can be defined on secret, job, folder or global.
    // the default is engine version 2 unless otherwise specified globally.
    def secrets = [
        [path: 'secret/testing', engineVersion: 1, secretValues: [
            [envVar: 'testing', vaultKey: 'value_one'],
            [envVar: 'testing_again', vaultKey: 'value_two']]],
        [path: 'secret/another_test', engineVersion: 2, secretValues: [
            [vaultKey: 'another_test']]]
    ]

    // optional configuration, if you do not provide this the next higher configuration
    // (e.g. folder or global) will be used
    def configuration = [vaultUrl: 'http://my-very-other-vault-url.com',
                         vaultCredentialId: 'my-vault-cred-id',
                         engineVersion: 1]
    // inside this block your credentials will be available as env variables
    withVault([configuration: configuration, vaultSecrets: secrets]) {
        sh 'echo $testing'
        sh 'echo $testing_again'
        sh 'echo $another_test'
    }
}

In the future we might migrate to a BuildStep instead of a BuildWrapper.

Use of dynamic credentials

You may also want to use dynamically allocated credentials:

import hudson.util.Secret
import com.cloudbees.plugins.credentials.CredentialsScope
import com.datapipe.jenkins.vault.credentials.VaultTokenCredential

VaultTokenCredential customCredential = new VaultTokenCredential(
    CredentialsScope.GLOBAL,
    'custom-credential',
    'My Custom Credential',
    Secret.fromString('This is my token. There are many like it, but this one is mine. My token is my best friend.')
)

node {
...
    def configuration = [vaultUrl: 'http://my-very-other-vault-url.com',
                         vaultCredential: customCredential]
...

Setting a vaultCredential will override any previously defined vaultCredentialId.

Works with any VaultCredential: VaultTokenCredential, VaultAppRoleCredential, etc.

Inject Vault Credentials into your Job

Pipeline Usage

withCredentials Block

Specify the variables for the vault address and token. Vault Address and Credentials are both required.
addrVariable and tokenVariable are optional. They will be set to VAULT_ADDR and VAULT_TOKEN respectively if omitted.

node {
    withCredentials([[$class: 'VaultTokenCredentialBinding', credentialsId: 'vaulttoken', vaultAddr: 'https://localhost:8200']]) {
        // values will be masked
        sh 'echo TOKEN=$VAULT_TOKEN'
        sh 'echo ADDR=$VAULT_ADDR'
    }
}

FreeStyle Job

freeStyle Credential Binding

Configuration as Code

There is an easier way to setup the global Vault configuration on your Jenkins server.
No need for messing around in the UI.

Jenkins Configuration as Code often shorten to JCasC or simplify Configuration as Code plugin allows you to configure Jenkins via a yaml file. If you are a first time user, you can learn more about JCasC 👈

Hashicorp Plugin also adds an extension to JCasC by providing a Secret Source for Configuration as Code plugin to read secrets from, which you can read about here

Prerequisite:

Install Configuration as Code Plugin on your Jenkins instance.

Refer to Installing a new plugin in Jenkins.

Add configuration YAML:

There are multiple ways to load JCasC yaml file to configure Jenkins:

  • JCasC by default searches for a file with the name jenkins.yaml in $JENKINS_ROOT.

  • The JCasC looks for an environment variable CASC_JENKINS_CONFIG which contains the path for the configuration yaml file.

    • A path to a folder containing a set of config files e.g. /var/jenkins_home/casc_configs.

    • A full path to a single file e.g. /var/jenkins_home/casc_configs/jenkins.yaml.

    • A URL pointing to a file served on the web e.g. https://<your-domain>/jenkins.yaml.

  • You can also set the configuration yaml path in the UI. Go to <your-jenkins-domain>/configuration-as-code. Enter path or URL to jenkins.yaml and select Apply New Configuration.

To configure your Vault in Jenkins add the following to jenkins.yaml:

unclassified:
  hashicorpVault:
    configuration:
      vaultCredentialId: "vaultToken"
      vaultUrl: "https://vault.company.io"

credentials:
  system:
    domainCredentials:
      - credentials:
          - vaultTokenCredential:
              description: "Uber Token"
              id: "vaultToken"
              scope: GLOBAL
              token: "${MY_SECRET_TOKEN}"

See handling secrets section in JCasC documentation for better security.

You can also configure VaultGithubTokenCredential, VaultGCPCredential, VaultAppRoleCredential or VaultAwsIamCredential.

If you are unsure about how to do it from yaml. You can still use the UI to configure credentials.
After you configured Credentials and the Global Vault configuration.
you can use the export feature build into JCasC by visiting <your-jenkins-domain>/configuration-as-code/viewExport

HashiCorp Vault Plugin as a Secret Source for JCasC

We can provide these initial secrets for JCasC. The secret source for JCasC is configured via environment variables as way to get access to vault at startup and when configuring Jenkins instance.

For Security and compatibility considerations please read more here

  • The environment variable CASC_VAULT_PW must be present, if token is not used and appRole/Secret is not used. (Vault password.)
  • The environment variable CASC_VAULT_USER must be present, if token is not used and appRole/Secret is not used. (Vault username.)
  • The environment variable CASC_VAULT_APPROLE must be present, if token is not used and U/P not used. (Vault AppRole ID.)
  • The environment variable CASC_VAULT_APPROLE_SECRET must be present, if token is not used and U/P not used. (Vault AppRole Secret ID.)
  • The environment variable CASC_VAULT_KUBERNETES_ROLE must be present, if you want to use Kubernetes Service Account. (Vault Kubernetes Role.)
  • The environment variable CASC_VAULT_AWS_IAM_ROLE must be present, if you want to use AWS IAM authentiation. (Vault AWS IAM Role.)
  • The environment variable CASC_VAULT_AWS_IAM_SERVER_ID must be present when using AWS IAM authentication and the Vault auth method requires a value for the X-Vault-AWS-IAM-Server-ID header. (Vault AWS IAM Server ID.)
  • The environment variable CASC_VAULT_TOKEN must be present, if U/P is not used. (Vault token.)
  • The environment variable CASC_VAULT_PATHS must be present. (Comma separated vault key paths. For example, secret/jenkins,secret/admin.)
  • The environment variable CASC_VAULT_URL must be present. (Vault url, including port number.)
  • The environment variable CASC_VAULT_AGENT_ADDR is optional. It takes precedence over CASC_VAULT_URL and is used for connecting to a Vault Agent. See this section
  • The environment variable CASC_VAULT_MOUNT is optional. (Vault auth mount. For example, ldap or another username & password authentication type, defaults to userpass.)
  • The environment variable CASC_VAULT_NAMESPACE is optional. If used, sets the Vault namespace for Enterprise Vaults.
  • The environment variable CASC_VAULT_PREFIX_PATH is optional. If used, allows to use complex prefix paths (for example with KV secrets available at my/long/data/prefix/kv/secret1 set this to my/long/data/prefix/kv).
  • The environment variable CASC_VAULT_FILE is optional, provides a way for the other variables to be read from a file instead of environment variables.
  • The environment variable CASC_VAULT_ENGINE_VERSION is optional. If unset, your vault path is assumed to be using kv version 2. If your vault path uses engine version 1, set this variable to 1.
  • The issued token should have read access to vault path auth/token/lookup-self in order to determine its expiration time. JCasC will re-issue a token if its expiration is reached (except for CASC_VAULT_TOKEN).

If the environment variables CASC_VAULT_URL and CASC_VAULT_PATHS are present, JCasC will try to gather initial secrets from Vault. However for it to work properly there is a need for authentication by either the combination of CASC_VAULT_USER and CASC_VAULT_PW, a CASC_VAULT_TOKEN, the combination of CASC_VAULT_APPROLE and CASC_VAULT_APPROLE_SECRET, a CASC_VAULT_KUBERNETES_ROLE, or a CASC_VAULT_AWS_IAM_ROLE. The authenticated user must have at least read access.

You can also provide a CASC_VAULT_FILE environment variable where you load the secrets from a file.

File should be in a Java Properties format

CASC_VAULT_PW=PASSWORD
CASC_VAULT_USER=USER
CASC_VAULT_TOKEN=TOKEN
CASC_VAULT_PATHS=secret/jenkins/master,secret/admin
CASC_VAULT_URL=https://vault.dot.com
CASC_VAULT_MOUNT=ldap

A good use for CASC_VAULT_FILE would be together with docker secrets.

version: "3.6"

services:
  jenkins:
    environment:
      CASC_VAULT_FILE: /run/secrets/jcasc_vault
    restart: always
    build: .
    image: jenkins.master:v1.0
    ports:
      - 8080:8080
      - 50000:50000
    volumes:
      - jenkins-home:/var/jenkins_home
    secrets:
      - jcasc_vault

volumes:
  jenkins-home:

secrets:
  jcasc_vault:
    file: ./secrets/jcasc_vault

Example: HashiCorp Vault Plugin as a Secret Source for JCasC

Assume Vault has a secret at path secret/jenkins/passwords with keys secretKey1 and secretKey2. Set the value for environment variable CASC_VAULT_PATHS to secret/jenkins/passwords. The Hashicorp Vault Plugin provides two ways of accessing the secrets: using just the key within the secret and using the full path to the secret key. The full path option allows for you to reference multiple secrets with overlapping keys.

credentials:
  system:
    domainCredentials:
      - credentials:
          - string:
              description: "Secret using only secret key name"
              id: "secretUsingKey"
              scope: GLOBAL
              token: "${secretKey1}"
          - string:
              description: "Secret using full path"
              id: "secretUsingKey"
              scope: GLOBAL
              token: "${secret/jenkins/passwords/secretKey1}"

Vault Agent

For the use-case of Vault Agent read here

When CASC_VAULT_AGENT_ADDR is specified, you only need to specify CASC_VAULT_PATHS and optionally CASC_VAULT_ENGINE_VERSION
Since Vault Agent must be configured to handle auto authentication.

Here is an example of how to configure your Vault Agent with an app role. Vault Agent supports multiple auto-auth methods

pid_file = "/tmp/agent_pidfile"
auto_auth {
    method {
        type = "approle"
        config = {
            role_id_file_path = "/home/vault/role_id"
            secret_id_file_path = "/home/vault/secret_id"
        }
    }
    sink {
        type = "file"
        config = {
            path = "/tmp/file-foo"
        }
    }
}
cache {
    use_auto_auth_token = true
}
listener "tcp" {
    address = "0.0.0.0:8200"
    tls_disable = true
}

Ideally your Vault Agent should be running on the same Machine or running as a Container networked together.
You ought to block any connection to Vault Agent for anything that is not considered localhost.

For setup with Jenkins and Vault Agent running on the same host you can achieve this by using

listener "tcp" {
    address = "127.0.0.1:8200"
    tls_disable = true
}

For Containers you would need to use listener address of 0.0.0.0:8200.
You should never expose the Vault Agent port. You OUGHT to network Vault Agent container and Jenkins container together.

listener "tcp" {
    address = "0.0.0.0:8200"
    tls_disable = true
}