CHANGES.txt

Hadoop EC2 Control Scripts Change Log

Trunk (unreleased)

  Create a unified PAC file for EC2 clusters.

  Enable map output compression by default.

  Add --auto-shutdown option for automatically shutting down instances after a
  given period (in minutes).

  Support both 0.18 and 0.20 clusters.

  Improve the proxy to return the PID of the proxy process. Typing

    eval `hadoop-ec2 proxy my-hadoop-cluster`

  starts the proxy and prints its PID. The PID is also set in the environment
  variable HADOOP_EC2_PROXY_PID.

  Add --client-cidr option which is used to set the firewall rules that allow
  the client to access the cluster.

  Add --force option to terminate-cluster and delete-storage.

  Fix bug in storage attachment code. (Contributed by Mark 'delduderino')

  Add support for --availability-zone option to create-storage.

0.2.0-beta

  Allow options to be specified on the command line, not just in a configuration
  file.

  Add --user-packages option for specifying user packages to install on
  instances.

  Unified remote instance boot script (hadoop-ec2-init-remote.sh) to support
  all combinations of OS (Fedora, Ubuntu) and storage (local disk, EBS).

  Make user_data_file configuration parameter optional.

  Make launch command block until Hadoop daemons are running.

  Generate local hadoop-site.xml to aid remote job submission.

  Use XFS filesystem for local disks.

0.1.1-beta

  Add authorization for port 80 access from client machine; and to master from
  itself. Also print master web server URL.

  Use standard Hadoop port numbers for namenode and jobtracker (8020, 8021).

  Generalized passing env variables to EC2 boot scripts.

  Add support for Ubuntu.

  Fix incorrect JSON spec in documentation.

0.1.0-beta

  First public release