Service to monitor Cloud Foundry applications.

Running applications in Cloud Foundry is nice, but the visibility of operational metrics leaves a little to be desired.

This application (which itself can be run as a worker application within a Cloud Foundry provider) polls Cloud Foundry APIs to get application metrics and events out of Cloud Foundry and put them into your collection system of choice.

Running

Usage: cf-metrics

ZONE_PREFIXES – the environment variable that contains a comma-separated list of
prefixes for ENV vars that can be used for authenticating with a Cloud Foundry
provider.

-verbose
    Enable verbose logging
-whitelist string
    A comma-separated list of app names to collect data about. If none
    specified, then defaults to all apps that the account can view.

For example:

> env ZONE_PREFIXES=PWS,CH [email protected] \
PWS_PASSWORD='some-passphrase' \
PWS_API=https://api.run.pivotal.io \
PWS_NAME=pivotal \
[email protected] \
CH_PASSWORD='some-other-passphrase' \
CH_API=https://api.lyra-836.appcloud.swisscom.com \
CH_NAME=swisscom cf-metrics

That will run cf-metrics, pointing at PWS and Swisscom public Cloud Foundry instances. It will poll every 10 seconds, checking all of the apps those accounts can see, for cpu and memory metrics along with application events.

For each application, these will be emitted as:

{
  "zone":"PWS",
  "space":"development",
  "app":"hello-python-web",
  "type":"metric",
  "timestamp":"2016-11-09T10:56:06.711449498Z",
  "stats":{
    "0":{
      "stats":{
        "diskQuota":536870912,
        "memQuota":402653184,
        "usage":{
          "cpu":0.006985241502010499,
          "disk":186060800,
          "mem":338554880,
          "diskUsage":0.346565247,
          "memUsage":0.84081014
        }
      }
    },
    "1":{
      "stats":{
        "diskQuota":536870912,
        "memQuota":402653184,
        "usage":{
          "cpu":0.007506122659531552,
          "disk":186060800,
          "mem":256016384,
          "diskUsage":0.34656524658203,
          "memUsage":0.63582356770833
        }
      }
    },
    "2":{
      "stats":{
        "diskQuota":536870912,
        "memQuota":402653184,
        "usage":{
          "cpu":0.004021443931726263,
          "disk":186056704,
          "mem":317755392,
          "diskUsage":0.3465576171875,
          "memUsage":0.78915405273438
        }
      }
    },
    "3":{
      "stats":{
        "diskQuota":536870912,
        "memQuota":402653184,
        "usage":{
          "cpu":0.00431447831668318,
          "disk":186073088,
          "mem":250630144,
          "diskUsage":0.34658813476563,
          "memUsage":0.62244669596354
        }
      }
    }
  }
}

Since these are point-in-time values, we can happily collect and aggregate those by running cf-metrics in multiple locations, and feed all of the logs into a single location.

It will also poll every 1 minute, looking for events that have happened for each application (such as process crash, restart) since cf-metrics began polling.

For each application, these events will be emitted as:

{
  "zone":"PWS",
  "space":"development",
  "app":"hello-python-web",
  "type":"event",
  "timestamp":"2016-11-09T11:49:53.575738294Z",
  "eventInfo":{
    "type":"app.crash",
    "timestamp":"2016-09-20T16:21:29Z"
  }
}

Cloud Foundry manifest

If you did want to run this application in Cloud Foundry, you could do that starting with a manifest like:

---
applications:
- name: worker-app
  buildpack: https://github.com/cloudfoundry/binary-buildpack.git
  command: chmod +x cf-metrics-linux-amd64; exec ./cf-metrics-linux-amd64 -whitelist app1,app2
  no-route: true
  health-check-type: none
  memory: 64M
  env:
    ZONE_PREFIXES: PWS,CH
    PWS_USERNAME: [email protected]
    PWS_PASSWORD: some-passphrase
    PWS_API: https://api.run.pivotal.io
    PWS_NAME: pivotal
    CH_USERNAME: [email protected]
    CH_PASSWORD: some-other-passphrase
    CH_API: https://api.lyra-836.appcloud.swisscom.com
    CH_NAME: swisscom

And deploy that using Concourse like:

jobs:
- name: deploy-to-prod
  plan:
  - get: gh-metrics-release
    version: { tag: v0.4.5 }
    params:
      globs:
      - cf-metrics-linux-amd64*
  - get: gh-repo
  - aggregate:
    - put: deploy-app-pws-prod
      params:
        current_app_name: {{APP_NAME}}
        manifest: gh-repo/infra/cloud-foundry/manifest.yml
        path: gh-metrics-release

resources:
- name: gh-repo
  type: git
  source:
    uri: {{GIT_REPO}}
    branch: master
    private_key: {{GIT_PRIVATE_KEY}}

- name: gh-metrics-release
  type: github-release
  source:
    user: jabley
    repository: cf-metrics
    token: {{GIT_ACCESS_TOKEN}}

- name: deploy-app-pws-prod
  type: cf-resource
  source:
    api: {{PWS_API}}
    username: {{PWS_USER}}
    password: {{PWS_PASS}}
    organization: {{PWS_ORG}}
    space: {{PWS_PROD_SPACE}}

Building

> make