… send replies to all of those commands. Fixes bug #12: https://github.com/starling/starling/issues/issue/12
…atible with the latest memcache-client gem. https://github.com/starling/starling/issues#issue/5
…ts officially released as memcache-client
If the config file doesn't contain a starling section or is not a hash exit with an error message indicating the problem instead of simply putting a backtrace in the users face.
* Client-side implementation ensures that the queue is deleted from every server in the cluster, since the Starling 'set' call will spread the items across all servers. Delegates to superclass for actual writing of delete request to socket, and manipulates #get_server_for_key to ensure that the queue is deleted from all servers. * When using the MemCache client, this isn't necessary since all queue items are put on the same Starling instance, so the 'delete' command is sent to that same server. * The server-side handler parses the standard MemCache delete message, so any MemCache client should be able to issue the command. * The rSpec test checks that the MemCache client can call the delete command, and also uses the Starling client to ensure that the number of available queues is appropriately updated. * Remove the queue's log file when deleting a queue. Otherwise the queue directory can get so full of log files (when frequently adding and removing queues) that the OS doesn't like it. Signed-off-by: Tim Shadel <firstname.lastname@example.org>
Simply use the memcache client superclass method, instead of the locally defined blocking version. Written by Aaron Hawkins <email@example.com>. Signed-off-by: Aaron Hawkins <firstname.lastname@example.org> Signed-off-by: Tim Shadel <email@example.com>
…or stats responses