如何用Puppet管理Ambari Server并调用其API启动服务？

Got it, this is a common friction point when moving from Ansible (which plays nice with imperative API calls) to Puppet's declarative, resource-focused model. The key is bridging that gap between Puppet's "state desired" approach and the imperative API calls needed to control Ambari services. Here are your best options:

1. Use exec Resources (Quickest Fix) #

Puppet's exec resource lets you run arbitrary commands, and you can add guard clauses (onlyif/unless) to ensure it only runs when the service isn't already in the desired state—keeping things idempotent, which is critical for Puppet.

Example: Start Hive via Ambari API with curl #

# Define Ambari credentials (store these in Hiera with encryption for production!)
$ambari_host = 'your-ambari-server.example.com'
$ambari_port = '8080'
$ambari_user = 'admin'
$ambari_pass = 'your-secure-password'
$cluster_name = 'your-cluster'

exec 'start_hive_service' {
  command => "curl -u ${ambari_user}:${ambari_pass} -X PUT -H 'X-Requested-By: ambari' http://${ambari_host}:${ambari_port}/api/v1/clusters/${cluster_name}/services/HIVE -d '{\"ServiceInfo\":{\"state\":\"STARTED\"}}'",
  path    => ['/usr/bin', '/bin'],
  # Only run if Hive is NOT already started
  onlyif  => "curl -s -u ${ambari_user}:${ambari_pass} http://${ambari_host}:${ambari_port}/api/v1/clusters/${cluster_name}/services/HIVE | grep -q '\"state\":\"INSTALLED\"'",
  # Allow time for Ambari to process the start request
  timeout => 300,
}

You can replicate this pattern for HDFS, SmartSense, or any other Ambari-managed service—just swap out the service name in the API endpoints and onlyif check.

2. Build a Custom Puppet Type/Provider (Cleanest, Most Puppet-Idiomatic) #

For a scalable, maintainable solution, wrap Ambari API interactions into a custom declarative resource. This lets you define service states just like you would with Puppet's native service resource.

Step 1: Define the Custom Type #

Save this in modules/ambari/lib/puppet/type/ambari_service.rb:

Puppet::Type.newtype(:ambari_service) do
  @doc = 'Manages the state of an Ambari-managed service'

  ensurable do
    newvalue(:running) do
      provider.start
    end
    newvalue(:stopped) do
      provider.stop
    end
    defaultto(:running)
  end

  newparam(:name) do
    desc 'The name of the Ambari service (e.g., HIVE, HDFS, SMARTSENSE)'
    isnamevar
  end

  newparam(:cluster_name) do
    desc 'Name of the Ambari cluster'
    isrequired
  end

  newparam(:ambari_host) do
    desc 'Ambari server hostname/IP'
    isrequired
  end

  newparam(:ambari_port) do
    desc 'Ambari server port'
    defaultto('8080')
  end

  newparam(:username) do
    desc 'Ambari admin username'
    isrequired
  end

  newparam(:password) do
    desc 'Ambari admin password'
    isrequired
  end
end

Step 2: Implement the Provider #

Save this in modules/ambari/lib/puppet/provider/ambari_service/ruby.rb:

require 'net/http'
require 'json'

Puppet::Type.type(:ambari_service).provide(:ruby) do
  def exists?
    # Check current service state via Ambari API
    uri = URI("http://#{resource[:ambari_host]}:#{resource[:ambari_port]}/api/v1/clusters/#{resource[:cluster_name]}/services/#{resource[:name]}")
    req = Net::HTTP::Get.new(uri)
    req.basic_auth(resource[:username], resource[:password])
    req['X-Requested-By'] = 'ambari'

    res = Net::HTTP.start(uri.hostname, uri.port) do |http|
      http.request(req)
    end

    state = JSON.parse(res.body)['ServiceInfo']['state']
    state == 'STARTED'
  end

  def start
    uri = URI("http://#{resource[:ambari_host]}:#{resource[:ambari_port]}/api/v1/clusters/#{resource[:cluster_name]}/services/#{resource[:name]}")
    req = Net::HTTP::Put.new(uri)
    req.basic_auth(resource[:username], resource[:password])
    req['X-Requested-By'] = 'ambari'
    req['Content-Type'] = 'application/json'
    req.body = { 'ServiceInfo' => { 'state' => 'STARTED' } }.to_json

    Net::HTTP.start(uri.hostname, uri.port) do |http|
      http.request(req)
    end
  end

  def stop
    # Similar to start, but set state to 'STOPPED'
    uri = URI("http://#{resource[:ambari_host]}:#{resource[:ambari_port]}/api/v1/clusters/#{resource[:cluster_name]}/services/#{resource[:name]}")
    req = Net::HTTP::Put.new(uri)
    req.basic_auth(resource[:username], resource[:password])
    req['X-Requested-By'] = 'ambari'
    req['Content-Type'] = 'application/json'
    req.body = { 'ServiceInfo' => { 'state' => 'STOPPED' } }.to_json

    Net::HTTP.start(uri.hostname, uri.port) do |http|
      http.request(req)
    end
  end
end

Step 3: Use the Custom Resource in Your Manifest #

ambari_service { 'HIVE':
  ensure        => running,
  cluster_name  => 'your-cluster',
  ambari_host   => 'your-ambari-server.example.com',
  username      => 'admin',
  password      => 'your-secure-password',
}

ambari_service { 'HDFS':
  ensure        => running,
  cluster_name  => 'your-cluster',
  ambari_host   => 'your-ambari-server.example.com',
  username      => 'admin',
  password      => 'your-secure-password',
}

This approach fits perfectly with Puppet's declarative model—you define the desired state, and the provider handles the API calls and state checks automatically.

3. Leverage Community Modules (If Available) #

Check the Puppet Forge for maintained Ambari modules (e.g., puppet-ambari). These often include pre-built resources for managing services out of the box, saving you from writing custom code. Just verify compatibility with your Ambari version before using.

Key Notes for All Solutions #

• Idempotency is Non-Negotiable: Always add checks (onlyif for exec, exists? for custom providers) to avoid repeating API calls unnecessarily.
• Secure Credentials: Never hardcode passwords in manifests. Use Hiera with encryption (e.g., eyaml) or Puppet Vault to store sensitive data.
• Timeout Handling: Ambari service starts can take several minutes—set appropriate timeouts to avoid Puppet failing prematurely.

内容的提问来源于stack exchange，提问作者L.HAMES