reliable-get

Reliable HTTP get wrapper (with cache and serve stale on error), best wrapped around things you dont trust very much.

Behaviour

If reliable-get makes a request that times out or errors, its callback will receive both an error object and a previously cached response, if one is present in cache. You can then decide whether to ignore the error and use the cached response, or not.

Basic usage

var ReliableGet = require('reliable-get');
var config = {
  cache:{
    engine:'memorycache'
  }
};
var rg = new ReliableGet(config);
rg.on('log', function(level, message, data) {
  // Wire up to your favourite logger
});
rg.on('stat', function(type, key, value) {
 // Wire up to your favourite stats library (e.g. statsd)
});
rg.get({url:'http://www.google.com'}, function(err, response) {
   console.log(response.content);
});

Configuration options

When you create an instance of reliable-get you need to specify the cache configuration. This then applies across all requests.

var config = {
  cache:{
    engine:'redis',
    url:'redis://localhost:6379?db=0'
  }
};

You can also pass a property requestOpts to pass options to be used in request. Example:

var config = {
  cache: {
    engine: 'redis',
    url: 'redis://localhost:6379?db=0'
  },
  requestOpts: {
    forever: true,
    followRedirect: false
  }
}

Property	Description	Example / Default	Required
cache.engine	Cache to use, redis/memorycache/nocache	nocache	No
cache.engine.url	URL to redis	localhost:6379	No
cache.compress	Use snappy compression	false	No
cache.namespace	Prefix for redis keys	''	No

GET options

When making a get request, you need to provide a basic options object:

rg.get({url:'http://www.google.com'}, function(err, response) {
   console.log(response.content);
});

Property	Description	Example / Default	Required
url	Service to get	http://my-service.tes.co.uk	Yes
timeout	Timeout for service	5000	No
cacheKey	Key to store cached value against	my-service_tes_co_uk	No
cacheTTL	TTL of cached value in ms	1 minute (60000)	No
explicitNoCache	Do not cache under any circumstances	false	No
headers	Headers to send with request		No
tracer	Unique value to pass with request		No
type	Type of request, used for statsd and logging		No
statsdKey	Key that statsd events will be posted to		No
eventHandler	Object (see below) for logging and stats		No

Example from a Compoxure backend request:

var options = {
          url: targetUrl,
          cacheKey: targetCacheKey,
          cacheTTL: targetCacheTTL,
          timeout: utils.timeToMillis(backend.timeout || DEFAULT_LOW_TIMEOUT),
          headers: backendHeaders,
          tracer: req.tracer,
          statsdKey: 'backend_' + utils.urlToCacheKey(host)
};

From a compoxure fragment request:

var options = {
      url: url,
      timeout: timeout,
      cacheKey: cacheKey,
      cacheTTL: cacheTTL,
      explicitNoCache: explicitNoCache,
      headers: optionsHeaders,
      tracer: req.tracer,
      statsdKey: statsdKey
  }

Configuration

Cache configuration

The cache object accept any config value accepted by redis. It also takes:

• config.cache.engine: nocache, memorycache, redis (use nocache/memorycache for testing only!)
• config.cache.url: redis dsn, it is translated to the connection parameters
• config.cache.compress: enable snappy compression on cached items
• config.cache.namespace: adds this string as a prefix to any key. Useful to share redis with other services or migrations

Request Configuration

The "config.requestOpts" contains the default configuration passed to "request".