I'm happy to announce the release of beanstalkd 0.5. This is the very first public release. In the future I'll confine release announcements to the beanstalk mailing list.
WHAT IS BEANSTALKD? ------------------- Beanstalkd is a fast, distributed, in-memory workqueue service. Its interface is generic, but is intended for use in reducing the latency of page views in high-volume web applications by running most time-consuming tasks asynchronously. Philotic, Inc. developed beanstalkd to improve the response time for the Causes on Facebook (http://apps.facebook.com/causes/) application (with over 9.5 million users). Beanstalkd drastically decreased the average response time for the most common pages to a tiny fraction of the original and dramatically improved the user experience. Stay tuned for a Rails plugin called Async Observer that will make using beanstalk in your Rails project a snap! OUR FEW URLS ------------ Download the 0.5 tarball and learn all about beanstalk: http://xph.us/software/beanstalkd/ Talk about beanstalk development or use at: http://groups.google.com/group/beanstalk-talk HOW TO RUN IT ------------- First, run beanstalkd on two or three machines. There is no configuration file and only a handful of command-line options. $ ./beanstalkd -d -l 10.0.1.5 -p 11300 This starts up beanstalkd as a daemon listening on address 10.0.1.5, port 11300. If you want to spy on what beanstalkd is doing, recompile it in debug mode. HOW TO USE IT ------------- Here's an example in Ruby. First, have one process put a job into the queue: beanstalk = Beanstalk::Pool(['localhost:11300']) ... beanstalk.put('hello') Then start another process to take jobs out of the queue and run them: beanstalk = Beanstalk::Pool(['localhost:11300']) loop do j = beanstalk.reserve puts j.body # prints "hello" j.delete end DEPENDENCY ---------- Beanstalkd requires libevent. It was developed with libevent 1.3b, but it may work with other versions. http://monkey.org/~provos/libevent/ KNOWN BUGS ---------- * The min heap data structure is not dynamically-sized. This means a fixed hard limit on the number of ready jobs (by default set to around 16 million). * The policy of signalling errors by closing the connection is wrong. The server should send a helpful error response whenever possible and only close the connection as a last resort. kr
