Permalink
Commits on Sep 28, 2009
  1. moving fire_callback into a deferred thread to avoid potential deadlo…

    jashkenas committed Sep 28, 2009
    …cks with a single server instance ... rescuing all rest-client exceptions in communication with nodes as evidence that the node has gone away
Commits on Sep 24, 2009
  1. heading off the potential for race conditions in multithreading the w…

    jashkenas committed Sep 24, 2009
    …ork unit dispatch by switching to secure random numbers, not process ID
Commits on Sep 23, 2009
  1. tiny tweaks

    jashkenas committed Sep 23, 2009
  2. adding a test that ensures that failed attempts to shell out in an Ac…

    jashkenas committed Sep 23, 2009
    …tion raise an exception (which in turns marks the WorkUnit as failed)
Commits on Sep 22, 2009
  1. removing a FloatDomainError from Job#percent_complete that never occu…

    jashkenas committed Sep 22, 2009
    …rs in practice, but you can force by mucking around in the console
Commits on Sep 20, 2009
Commits on Sep 19, 2009
  1. can't put backticks in a commit message -- the last two commits added…

    jashkenas committed Sep 19, 2009
    … daemonizing support for nodes and servers through 'crowd --daemonize'
  2. updating

    jashkenas committed Sep 19, 2009
    CloudCrowd is a MapReduce-inspired Parallel Processing System for Ruby.
    
    Wiki: http://wiki.github.com/documentcloud/cloud-crowd
    Rdoc: http://rdoc.info/projects/documentcloud/cloud-crowd
    
    Usage: crowd COMMAND OPTIONS
    
    Commands:
      install       Install the CloudCrowd configuration files to the specified directory
      server        Start up the central server (requires a database)
      node          Start up a worker node (only one node per machine, please)
      console       Launch a CloudCrowd console, connected to the central database
      load_schema   Load the schema into the database specified by database.yml
    
    Options:
        -c, --config PATH                path to configuration directory
        -p, --port PORT                  port number for server (central or node)
        -e, --environment ENV            server environment (sinatra)
        -v, --version                    show version help with daemon options
  3. CloudCrowd is a MapReduce-inspired Parallel Processing System for Ruby.

    jashkenas committed Sep 19, 2009
    Wiki: http://wiki.github.com/documentcloud/cloud-crowd
    Rdoc: http://rdoc.info/projects/documentcloud/cloud-crowd
    
    Usage: crowd COMMAND OPTIONS
    
    Commands:
      install       Install the CloudCrowd configuration files to the specified directory
      server        Start up the central server (requires a database)
      node          Start up a worker node (only one node per machine, please)
      console       Launch a CloudCrowd console, connected to the central database
      load_schema   Load the schema into the database specified by database.yml
    
    Options:
        -c, --config PATH                path to configuration directory
        -p, --port PORT                  port number for server (central or node)
        -e, --environment ENV            server environment (sinatra)
        -v, --version                    show version now knows how to daemonize servers and nodes -- options to configure log folder and pid folder added to config.yml
  4. changing brower-side absolute URLs to relative ones -- so that you ca…

    jashkenas committed Sep 19, 2009
    …n deploy CloudCrowd to sub-URIs, configure it in config.ru and config.yml, and be off to the races.
Commits on Sep 18, 2009
  1. version 0.2.1 is on the books

    jashkenas committed Sep 18, 2009
  2. added node record tests

    jashkenas committed Sep 18, 2009
  3. adding overloaded node tests

    jashkenas committed Sep 18, 2009
  4. adding a new worker-limiting option to accompany 'max_workers' and 'm…

    jashkenas committed Sep 18, 2009
    …ax_load': 'min_free_memory'
  5. adding one of the items on the wish list -- max_load, the maximum loa…

    jashkenas committed Sep 18, 2009
    …d average that a node is allowed to reach before refusing new work, temporarily
Commits on Sep 17, 2009
  1. added some robustness to worker management with a custom backticks im…

    jashkenas committed Sep 17, 2009
    …plementation in Action that raises an Error::CommandFailed when exit codes are non-zero
  2. moving the default test database to sqlite3, adding a commented out s…

    jashkenas committed Sep 17, 2009
    …qlite3 config to database.example.yml, and removing the (now unneeded) rake db:create_test_database task (thanks delagoya)