Class MemCache
In: lib/active_support/vendor/memcache-client-1.7.4/memcache.rb
Parent: Object

A Ruby client library for memcached.

Methods

Classes and Modules

Class MemCache::MemCacheError
Class MemCache::Server

Constants

VERSION = '1.7.4'   The version of MemCache you are using.
DEFAULT_OPTIONS = { :namespace => nil, :readonly => false, :multithread => true, :failover => true, :timeout => 0.5, :logger => nil, :no_reply => false, }   Default options for the cache object.
DEFAULT_PORT = 11211   Default memcached port.
DEFAULT_WEIGHT = 1   Default memcached server weight.
ONE_MB = 1024 * 1024   Add key to the cache with value value that expires in expiry seconds. If raw is true, value will not be Marshalled.

Warning: Readers should not call this method in the event of a cache miss; see MemCache#add.

Attributes

failover  [R]  Should the client try to failover to another server if the first server is down? Defaults to true.
logger  [R]  Log debug/info/warn/error to the given Logger, defaults to nil.
multithread  [R]  The multithread setting for this instance
namespace  [R]  The namespace for this instance
no_reply  [R]  Don‘t send or look for a reply from the memcached server for write operations. Please note this feature only works in memcached 1.2.5 and later. Earlier versions will reply with "ERROR".
servers  [R]  The servers this client talks to. Play at your own peril.
timeout  [R]  Socket timeout limit with this client, defaults to 0.5 sec. Set to nil to disable timeouts.

Public Class methods

Accepts a list of servers and a list of opts. servers may be omitted. See +servers=+ for acceptable server list arguments.

Valid options for opts are:

  [:namespace]   Prepends this value to all keys added or retrieved.
  [:readonly]    Raises an exception on cache writes when true.
  [:multithread] Wraps cache access in a Mutex for thread safety. Defaults to true.
  [:failover]    Should the client try to failover to another server if the
                 first server is down?  Defaults to true.
  [:timeout]     Time to use as the socket read timeout.  Defaults to 0.5 sec,
                 set to nil to disable timeouts (this is a major performance penalty in Ruby 1.8,
                 "gem install SystemTimer' to remove most of the penalty).
  [:logger]      Logger to use for info/debug output, defaults to nil
  [:no_reply]    Don't bother looking for a reply for write operations (i.e. they
                 become 'fire and forget'), memcached 1.2.5 and later only, speeds up
                 set/add/delete/incr/decr significantly.

Other options are ignored.

Public Instance methods

[](key, raw = false)

Alias for get

Shortcut to save a value in the cache. This method does not set an expiration on the entry. Use set to specify an explicit expiry.

Returns whether there is at least one active server for the object.

Add key to the cache with value value that expires in expiry seconds, but only if key does not already exist in the cache. If raw is true, value will not be Marshalled.

Readers should call this method in the event of a cache miss, not MemCache#set.

Append - ‘add this data to an existing key after existing data’ Please note the value is always passed to memcached as raw since it doesn‘t make a lot of sense to concatenate marshalled data together.

"cas" is a check and set operation which means "store this data but only if no one else has updated since I last fetched it." This can be used as a form of optimistic locking.

Works in block form like so:

  cache.cas('some-key') do |value|
    value + 1
  end

Returns: nil if the value was not found on the memcached server. STORED if the value was updated successfully EXISTS if the value was updated by someone else since last fetch

Decrements the value for key by amount and returns the new value. key must already exist. If key is not an integer, it is assumed to be

  1. key can not be decremented below 0.

Removes key from the cache in expiry seconds.

Performs a get with the given key. If the value does not exist and a block was given, the block will be called and the result saved via add.

If you do not provide a block, using this method is the same as using get.

Flush the cache from all memcache servers. A non-zero value for delay will ensure that the flush is propogated slowly through your memcached server farm. The Nth server will be flushed N*delay seconds from now, asynchronously so this method returns quickly. This prevents a huge database spike due to a total flush all at once.

Retrieves key from memcache. If raw is false, the value will be unmarshalled.

Retrieves multiple values from memcached in parallel, if possible.

The memcached protocol supports the ability to retrieve multiple keys in a single request. Pass in an array of keys to this method and it will:

  1. map the key to the appropriate memcached server
  2. send a single request to each server that has one or more key values

Returns a hash of values.

  cache["a"] = 1
  cache["b"] = 2
  cache.get_multi "a", "b" # => { "a" => 1, "b" => 2 }

Note that get_multi assumes the values are marshalled.

Increments the value for key by amount and returns the new value. key must already exist. If key is not an integer, it is assumed to be 0.

Returns a string representation of the cache object.

Prepend - ‘add this data to an existing key before existing data’ Please note the value is always passed to memcached as raw since it doesn‘t make a lot of sense to concatenate marshalled data together.

Returns whether or not the cache object was created read only.

Add key to the cache with value value that expires in expiry seconds, but only if key already exists in the cache. If raw is true, value will not be Marshalled.

Reset the connection to all memcache servers. This should be called if there is a problem with a cache lookup that might have left the connection in a corrupted state.

Set the servers that the requests will be distributed between. Entries can be either strings of the form "hostname:port" or "hostname:port:weight" or MemCache::Server objects.

Returns statistics for each memcached server. An explanation of the statistics can be found in the memcached docs:

code.sixapart.com/svn/memcached/trunk/server/doc/protocol.txt

Example:

  >> pp CACHE.stats
  {"localhost:11211"=>
    {"bytes"=>4718,
     "pid"=>20188,
     "connection_structures"=>4,
     "time"=>1162278121,
     "pointer_size"=>32,
     "limit_maxbytes"=>67108864,
     "cmd_get"=>14532,
     "version"=>"1.2.0",
     "bytes_written"=>432583,
     "cmd_set"=>32,
     "get_misses"=>0,
     "total_connections"=>19,
     "curr_connections"=>3,
     "curr_items"=>4,
     "uptime"=>1557,
     "get_hits"=>14532,
     "total_items"=>32,
     "rusage_system"=>0.313952,
     "rusage_user"=>0.119981,
     "bytes_read"=>190619}}
  => nil

Protected Instance methods

Performs a raw decr for cache_key from server. Returns nil if not found.

Fetches the raw data for cache_key from server. Returns nil on cache miss.

Fetches cache_keys from server using a multi-get.

Performs a raw incr for cache_key from server. Returns nil if not found.

Pick a server to handle the request based on a hash of the key.

Handles error from server.

Returns an interoperable hash value for key. (I think, docs are sketchy for down servers).

Create a key for the cache, incorporating the namespace qualifier if requested.

Performs setup for making a request with key from memcached. Returns the server to fetch the key from and the complete key to use.

Gets or creates a socket connected to the given server, and yields it to the block, wrapped in a mutex synchronization if @multithread is true.

If a socket error (SocketError, SystemCallError, IOError) or protocol error (MemCacheError) is raised by the block, closes the socket, attempts to connect again, and retries the block (once). If an error is again raised, reraises it as MemCacheError.

If unable to connect to the server (or if in the reconnect wait period), raises MemCacheError. Note that the socket connect code marks a server dead for a timeout period, so retrying does not apply to connection attempt failures (but does still apply to unexpectedly lost connections etc.).

[Validate]