CF Light API

A super lightweight API for CloudFoundry. Why? Well the CF API contains all sorts of relations making it very heavy to query.

Having lots of API consumers for random scripts and dashboards makes it necessary to scale up the CF installation not to disrupt normal "cf cli operations". So, lets just cache all the data we want in Redis for 5 minutes and serve that. \o/

This gem provides a single binary cf_light_api, which starts a small Sinatra app to serve the HTTP requests, and also starts a background worker via the Rufus Scheduler, which updates Redis with the data from CF every 5 minutes.

The API just reads a stringified JSON from Redis and serves it under the following endpoints.

There are currently two endpoints, the original (and now deprecated) "v1" and a new "v2" which offers more details and better error handling.

GET /v2/apps

GET /v2/apps/<org_name>

An array of JSON documents for all applications in the configured CF environment. If you provide an org parameter, the list will be filtered to contain only applications belonging to the given org. The majority of the document is identical to the response you would receive from a standard /v2/apps request made against the main CF API, but with some additional fields added on top:

• created_at - When the app was originally created, taken from the metadata.

• updated_at - When the app was last modified, taken from the metadata.

• guid - The app GUID, taken from the metadata.

• instances - An array of all instances associated with this application, their state and usage statistics

• routes - An array of all routes assigned to your app instances (also exposed via the uris attribute of each member of the instances attribute).

• meta - A document containing any error messages capturing during processing of this application.

• stack - The name of the container stack being used by the application, as a string.

• space - The name of the space this app belongs to, as a string.

• org - The name of the org this app belongs to, as a string.

• environment_json - An array of environment variables exposed to this application, if enabled as documented here.

• buildpack_name - A field containing the name of the buildpack in use. Its value is equal either to the buildpack attribute of the associated app or should this be none (autodetected buildpack) to the buildpack name as resolved from the v2/buildpacks/<guid> endpoint using the detected_buildpack_guid field as guid.

Each document has the following structure:

{
  "name": "app_name",
  "production": false,
  "space_guid": "c0af44b8-8b51-4db5-927e-ccad2e6dab54",
  "stack_guid": "7c5b664c-e9b8-457b-83a4-6092b7372494",
  "buildpack": null,
  "detected_buildpack": "Ruby",
  "detected_buildpack_guid": "44ec3a97-0d94-4ebb-ad33-e9ee837515bd",
  "environment_json": {},
  "memory": 64,
  "instances": [
    {
      "state": "RUNNING",
      "stats": {
        "name": "app_name",
        "uris": [
          "app_name.yourdomain.com"
        ],
        "host": "ip address",
        "port": "port number",
        "uptime": 2889203,
        "mem_quota": 67108864,
        "disk_quota": 1073741824,
        "fds_quota": 16384,
        "usage": {
          "time": "2018-01-19T15:54:48.254719257Z",
          "cpu": 0.00012013292649106141,
          "mem": 43802624,
          "disk": 78118912
        }
      }
    }
  ],
  "disk_quota": 1024,
  "state": "STARTED",
  "version": "45c81548-3d3d-4152-bbc5-6b86eba4d5df",
  "command": null,
  "console": false,
  "debug": null,
  "staging_task_id": "ba623c5c-18e1-4d6e-b331-aedf244cb493",
  "package_state": "STAGED",
  "health_check_type": "port",
  "health_check_timeout": null,
  "health_check_http_endpoint": null,
  "staging_failed_reason": null,
  "staging_failed_description": null,
  "diego": true,
  "docker_image": null,
  "docker_credentials": {
    "username": null,
    "password": null
  },
  "package_updated_at": "2017-06-13T12:19:37Z",
  "detected_start_command": "bundle exec rackup config.ru -p $PORT",
  "enable_ssh": false,
  "ports": [
    8080
  ],
  "space_url": "/v2/spaces/c0af44b8-8b51-4db5-927e-ccad2e6dab54",
  "stack_url": "/v2/stacks/7c5b664c-e9b8-457b-83a4-6092b7372494",
  "stack": "cflinuxfs2",
  "routes_url": "/v2/apps/ba623c5c-18e1-4d6e-b331-aedf244cb493/routes",
  "routes": [
    "app_name.yourdomain.com"
  ],
  "events_url": "/v2/apps/ba623c5c-18e1-4d6e-b331-aedf244cb493/events",
  "service_bindings_url": "/v2/apps/ba623c5c-18e1-4d6e-b331-aedf244cb493/service_bindings",
  "route_mappings_url": "/v2/apps/ba623c5c-18e1-4d6e-b331-aedf244cb493/route_mappings",
  "created_at": "2015-04-22T12:07:56Z",
  "updated_at": "2016-12-20T10:02:47Z",
  "guid": "ba623c5c-18e1-4d6e-b331-aedf244cb493",
  "running": true,
  "environment_variables": {},
  "meta": {
    "error": false
  },
  "buildpack_name": "buildpack_name",
  "space": "space name",
  "org": "org name"
}

If there are any errors processing a given application, details on the error will be captured and exposed in the meta attribute, for example:

"meta": {
  "error": true,
  "type": "CFResponseError",
  "message": "Code 200003: CF-AppStoppedStatsError - Could not fetch stats for stopped app: <name of app here>",
  "backtrace": [
    "etc...", "etc...", "etc..."
  ]
}

If there are no errors, the meta attribute will instead look like this:

"meta": {
  "error": false
}

• The running attribute is a boolean and will be true if there is at least one instance of your app with the state of RUNNING, or false otherwise.

• Memory, disk quota and usage figures are given in bytes.

• If the buildpack is not known, the buildpack attribute will be null. However, the buildpack_name field will be populated using the buildpack detected_buildpack_guid value.

• If the last uploaded time is not known, the last_uploaded attribute will be null.

• The diego attribute is a boolean which will be true if the application is running on a Diego Cell, or false if running on a DEA Node. See the Cloud Foundry docs for more information on these two architectures.

• There is an optional environment_json attribute which only appears if the feature is enabled - please see the Gathering Environment Variables section below for more information.

GET /v1/apps

GET /v1/apps/<org_name>

An array of JSON documents for all applications in the configured CF environment. If you provide an org parameter, the list will be filtered to contain only applications belonging to the given org. Each document has the following structure:

{
  "guid": "app GUID",
  "name": "app_name",
  "org": "org name",
  "space": "space name",
  "stack": "lucid64",
  "buildpack": "https://github.com/cloudfoundry/ruby-buildpack.git",
  "diego": false,
  "docker": false,
  "docker_image": null,
  "routes": [
    "app_name.yourdomain.com"
  ],
  "data_from": "unix timestamp of last update",
  "last_uploaded": "2016-02-04 15:21:25 +0000",
  "state": "STARTED",
  "running": true,
  "instances": [
    {
      "state": "RUNNING",
      "stats": {
        "name": "app_name",
        "uris": [
          "app_name.yourdomain.com"
        ],
        "host": "ip address",
        "port": "port number",
        "uptime": 1979434,
        "mem_quota": 268435456,
        "disk_quota": 268435456,
        "fds_quota": 16384,
        "usage": {
          "time": "2015-02-27 16:52:35 +0000",
          "cpu": 0.0,
          "mem": 134217728,
          "disk": 134217728
        }
      }
    }
  ],
  "error": null
}

• The running attribute may contain true, false or error. Applications in the latter state will have further information about the problem in the error attribute, which is null at all other times.

• Memory, disk quota and usage figures are given in bytes.

• If the buildpack is not known, the buildpack attribute will be null.

• If the last uploaded time is not known, the last_uploaded attribute will be null.

• The diego attribute is a boolean which will be true if the application is running on a Diego Cell, or false if running on a DEA Node. See the Cloud Foundry docs for more information on these two architectures.

• The docker attribute is a boolean which will be true if the application was deployed from a Docker image, or false if it was deployed using a buildpack. If true, the docker_image attribute will then show the Docker image used, otherwise it will be null.

• There is an optional environment_variables attribute which only appears if the feature is enabled - please see the section of Gathering Environment Variables below for more information.

GET /v1/orgs

GET /v1/orgs/<guid>

An array of JSON documents, for all organisations in your CF environment. If you provide an guid parameter, the list will be filtered to contain only the organisation with the given guid. Each document has the following structure:

{
  "guid": "org GUID",
  "name": "my_org_name",
  "quota": {
    "total_services": 50,
    "memory_limit": 10737418240
  }
}

• Memory limits for each org are given in bytes.

GET /v1/last_updated

A single JSON document showing the last time the data was updated by the worker:

{
  "last_updated": "2015-11-07 01:25:28 +0000"
}

If the data was not updated within a configurable time period, the HTTP status will be 503 Service Unavailable.

GET /v1/info

A single JSON document showing various pieces of information about the CF installation. This data comes unmodified from CF's own /v2/info endpoint.

{
  "name": "vcap",
  "build": "2222",
  "support": "http://support.cloudfoundry.com",
  "version": 2,
  "description": "Cloud Foundry sponsored by Pivotal",
  "authorization_endpoint": "http://localhost:8080/uaa",
  "token_endpoint": "http://localhost:8080/uaa",
  "min_cli_version": "6.20.0",
  "min_recommended_cli_version": "6.21.0",
  "api_version": "2.94.0",
  "app_ssh_endpoint": "ssh.system.domain.example.com:2222",
  "app_ssh_host_key_fingerprint": "47:0d:d1:c8:c3:3d:0a:36:d1:49:2f:f2:90:27:31:d0",
  "app_ssh_oauth_client": null,
  "routing_endpoint": "http://localhost:3000",
  "doppler_logging_endpoint": "wss://doppler.vcap.me:4443",
  "logging_endpoint": "ws://loggregator.vcap.me:80"
}

The worker basically gets all the data we want from the real API every 5 mins, puts in Redis and sleeps. The worker runs in a background thread via the Rufus Scheduler and is automatically started as part of the API. There is basic locking implemented via Redis to allow you to run multiple instances of cf_light_api alongside one another, without duplicating work.

1. You must first set the following environment variables:

  export REDIS_URI=redis://redis.yourdomain.com:6379/
  export REDIS_KEY_PREFIX=cf_light_api_live  #useful if you are sharing a single Redis database)
  export CF_API=https://api.cf.yourdomain.com
  export CF_USER=username
  export CF_PASSWORD=password

2. In a new Ruby project, create a Gemfile containing the following:

  ruby '2.0.0'
  source 'https://rubygems.org'

  gem 'cf_light_api'

Then run bundle install.

3. You should now be able to start the CF Light API and worker by running cf_light_api.

1. Create a manifest.yml in the Ruby project you just created, containing the following:

---
applications:
  - name: cf-light-api
    instances: 2
    memory: 128MB
    env:
      REDIS_URI: redis://redis.yourdomain.com:6379/
      REDIS_KEY_PREFIX: cf_light_api_live
      CF_API: https://api.cf.yourdomain.com
      CF_USER: username
      CF_PASSWORD: password

2. Then simply cf push when logged in to your CF environment.

You can enable more verbose debug logs by setting the following environment variable:

export DEBUG=true

The default is to update data every 5 minutes, with a 5 minute timeout. You can modify this behaviour by setting the following environment variables:

export UPDATE_INTERVAL='10m' export UPDATE_TIMEOUT='7m'

Any option which is valid for Rufus Scheduler durations will work here, for example 30s, 1m or 1d.

If you change these settings, you will also need to adjust the data age validity, as described below.

For larger Cloud Foundry deployments, you may wish to increase the number of threads (from the default of 1) used to gather application data by setting the following environment variable:

export UPDATE_THREADS=4

Note that this will of course increase the number of simultaneous requests made to your main Cloud Foundry API.

By default, the data is considered valid if it was last updated within 10 minutes (twice the default update interval). You can modify this behaviour by setting the following environment variable (in seconds):

export DATA_AGE_VALIDITY=3600

Usage statistics for each app instance and org quota details can be exported to Graphite by setting the following environment variables:

export GRAPHITE_HOST=graphiteserver.domain.com export GRAPHITE_PORT=2003 export CF_ENV_NAME=live

If you have specified the The Graphite schema will look like this:

"cf_apps.#{cf_env_name}.#{org_name}.#{space_name}.#{app_name}.#{app_instance_index}.#{cpu|mem|disk|mem_quota|disk_quota}"
"cf_orgs.#{cf_env_name}.#{org_name}.quota.#{total_services|total_routes|memory_limit}"

Instrumentation regarding the API and the backend worker can be made available in New Relic if you enable this integration by setting the following environment variables:

export NEW_RELIC_LICENSE_KEY=<license key goes here>
export NEW_RELIC_APP_NAME="CF Light API"

You can gather all user-provided environment variables for each application by enabling the feature with an environment variable:

export EXPOSE_ENVIRONMENT_VARIABLES=true

If you want to filter out sensitive environment variables, you can also set an optional whitelist, to capture only the environment variables you specify via a comma-seperated list of one or more strings:

export ENVIRONMENT_VARIABLES_WHITELIST="MY_ENV_VAR, A_SECOND_ENV_VAR"

The matching data will be exposed in an environment_variables attribute in the response to /v1/apps. If this feature is not explicitly enabled, the attribute will not be present at all. If you enable the feature with a whitelist and there are no matches, this attribute will contain an empty document {}.

Source hosted at GitHub. Report issues and feature requests on GitHub Issues.

• Fork the project.
• Make your feature addition or bug fix, ideally committing to a topic branch.
• Please document your changes in the README, but don't change the version, or changelog.
• Send a pull request.

Copyright (c) Springer Nature. See LICENSE for details.