HTTP监控和分析应用PingPong.zip

<p align="center"> <img src="public/pingpong-og.png" alt="Pingpong Logo"> </p> <img src="https://travis-ci.org/keen/pingpong.png?branch=master&foo=bar" alt="Pingpong Build Status"> [](https://heroku.com/deploy) #### Get deeper HTTP request response analytics every second. Track real-time performance and availability across multiple API servers to see the what, when, and how behind your system performance. So you can understand why.  #### How does it work? + Pingpong sends HTTP requests to URLs you configure as frequently as once per second. It turns data about each request and response into JSON, then logs it to a custom destination. + Your default data store is Keen IO’s [analytics API](https://keen.io/docs/) to capture events, run queries, and create visualizations. But it’s simple to set up another backend. + Pingpong ships with [Dashboards](http://keen.github.io/dashboards/), an HTML visualization kit that lets you see and arrange your most critical response data. Built on the Keen IO analytics API, Dashboards is super-flexible and ready to be skinned, tweaked, and embedded anywhere. + Pingpong captures most of the data you'd want about HTTP requests and responses. To beef up or slim down your data stream, adding custom properties specific to your infrastructure is simple. #### Choose your own install adventure. **Deploy straight to Heroku:** Pingpong is easy to install and ready for deployment to one or more Heroku regions. You can even deploy the app with a single click with this handy button: [](https://heroku.com/deploy) *A note on event limits:* If you're using the Keen IO backend to store events, you can send 50,000 events for free per month. As a reference, one check running every minute will create about 43,000 events in a month. Check out [more plans](https://keen.io/pricing) to get more events. We'd also love to give you a discount if you're using Pingpong, just [email us](mailto:team@keen.io?subject=Pingpong Events) your project ID and we'll get you hooked up. **Setup and deploy your own Pingpong app:** Don't run Heroku? That's cool. You can run Pingpong on any host with Ruby, even your local machine. Either way, it's up and running in less than five minutes. Just see the next section. #### Setup & Deployment Pingpong is open source and easy to install. Pingpong is written in Ruby and streamlined for deployment to one or more Heroku regions. That said, you can run it on any host with Ruby, including your local machine. **Step 1:** Clone or fork this repository: ``` $ git clone git@github.com:keen/pingpong.git $ cd pingpong ``` **Step 2:** Install dependencies ``` $ bundle install ``` If you don't have the `bundle` command, first `gem install bundler`. **Step 3:** Set up database tables ``` $ bundle exec rake db:create $ bundle exec rake db:migrate ``` The project uses Postgres as the default database, but you can modify that in the `database.yml` file. **Step 4:** Set up the environment variables ***Keen IO Setup*** You'll need to sign up for a free [Keen IO](https://keen.io?s=pingpong) account. Once your account is set up, create a new project. You'll need to grab the `project id`, `read key`, and `write key`. Add these to a root level file called `.env`. There's a .env.sample setup with all the variables, so just run: ```sh $ cp .env.sample .env ``` And edit your Keen IO variables: ``` KEEN_PROJECT_ID=xxxxxxxxxxxxxxx KEEN_READ_KEY=yyyyyyyyyyyyyyyyy KEEN_WRITE_KEY=zzzzzzzzzzzzzzzz ``` ***SendGrid Setup*** If you want to send emails, you'll have to sign up for a free [SendGrid](https://sendgrid.com/user/signup) account. Once you've done that, edit the following in your `.env` file: ``` TO_EMAIL_ADDRESS=me@test.com FROM_EMAIL_ADDRESS=pingpong@test.com SKIP_EMAIL=false SENDGRID_USERNAME=asdfasdfasdf SENDGRID_PASSWORD=12345 ``` ***Slack Setup*** To get notifications in Slack, you'll have to provide us with your Incoming Webhook URL. You can create an Incoming Webhook [here](https://slack.com/services/new/incoming-webhook). Once you've got the Webhook URL, update your `.env` file: ``` SLACK_WEBHOOK_URL=https://hooks.slack.com/services/XXXXXXXXXXXXXXX/YYYYYYYYYYY ``` There are also some optional configurations you can edit for Slack notifications: ``` SLACK_CHANNEL='#alerts' # The channel to send notifications to SLACK_USERNAME='Robot' # The username the notification will come from - defaults to Pingpong SLACK_ICON=':rotating_light' # The icon of the user "sending" the notification. Can be a URL or an Emoji WARN_COLOR='#CCCCCC' # Hex color value used for the warning messages - defaults to #E2E541 BAD_COLOR='#000000' # Hex color value used for the failure messages - defaults to #F25656 ``` ***Run the Server*** Now you're ready to start the web server locally using `foreman`, which will pick up the variables in the `.env` file. [foreman](https://github.com/ddollar/foreman) comes with the [Heroku toolbelt](https://toolbelt.heroku.com/). ``` $ foreman start ``` The Pingpong web interface should now be running on [localhost:5000](http://localhost:5000). Click on the button to create a new check, and then within a few minutes, you'll see the check data populating the charts. #### Check Properties Every check requires the following properties: + name: for display in charts and reports + url: the fully qualified resource to check + frequency: how often to sent the request, in minutes Additionally, checks have some optional properties: + method: GET, POST, or DELETE (defaults to GET) + http_username: Username for HTTP authentication + http_password: Password for HTTP authentication Checks can also have any number of custom properties, which is very useful for grouping & drill-down analysis later. Place any custom properties in the `custom` field. #### HTTP Request & Response as an Event Each time a check is run, a JSON object describing the check, request, and response is logged via a `CheckLogger` component, defaulting to `KeenCheckLogger`. Here's an example event payload: ``` json { "check": { "name": "Keen IO Web", "url": "https://keen.io", "frequency": 5, "custom": { "server_role": "https", "is_https": true } }, "environment": { "rack_env": "production", "region": "heroku_us_east", "location": "Virginia, US" }, "request": { "sent_at": "2013-10-12T00:00:00.000Z", "duration": 0.432 }, "response": { "successful": true, "timed_out": false, "status": 200, "server": "TornadoServer/3.1", "http_status": 200, "http_reason": "OK", "http_version": "1.1", "content_type": "text/html", "content_length": 175, "date": "Wed, 16 Apr 2014 17:39:01 GMT" } } ``` Here's a breakdown of the major sections: + check: properties describing the check, including any custom properties + environment: properties describing where the check was made from (useful when you are running Pingpong instances across multiple datacenters) + request: information about the HTTP request that was sent + response: information about the HTTP response that was recorded `response.timed_out` and `response.successful` are helper properties. `response.successful` is true if the request did not timeout and the response status is between 100 and 399. It's easy to add more fields to the `environment` section in `config.yml`, or implement a `CheckMarshaller` component that translates HTTP response fields to properties in a different way. Capturing all of these fields makes it possible to perform powerful grouping and filtering during analysis. #### Reporting and Alerting Pingpong uses [Pushpop](https://github.com/pushpop/pingpong.git) to provide basic alerting and reporting functionality. This functionality can easily be extended to create your own custom ale