No description or website provided.
Python Shell PHP
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


First things first

You need Cassandra running. See the DataStax Community Edition guide. Install apache-cassandra1 (rather than dsc) to only get the open-source Cassandra tools, which are all you need.

Initial setup on RHEL and CentOS (tested on CentOS 5)

  1. Install Python 2.6 and its setuptools:

    yum install -y python26 python26-distribute python26-setuptools python26-devel git gcc
    easy_install-2.6 pip
    easy_install-2.6 -U distribute
    pip-2.6 install git+
    ln -s /usr/bin/pip-2.6 /usr/bin/pip  # Optional, fixes Puppet PIP support.
  2. Start and enable the Cassandra and perfbucker-watcher services:

    chkconfig cassandra on
    chkconfig perfbucket-watcher on
    /etc/init.d/cassandra start
    /etc/init.d/perfbucket-watcher start

Initial setup on Fedora (tested on Fedora 16)

  1. Set up the Python side:

    yum install -y python-pip gcc python-devel git
    pip-python install git+
    ln -s /usr/bin/pip-python /usr/bin/pip  # Optional, fixes Puppet PIP support.
  2. Optional: Copy the bundled perfbucket-watcher.service file into /etc/systemd/system to override the init script.

  3. Start and enable the Cassandra and perfbucker-watcher services:

    systemctl enable cassandra.service perfbucket-watcher.service
    systemctl start cassandra.service perfbucket-watcher.service

Final setup for all Red Hat-style systems

  1. Initialize the perfbucket schema:

    perfbucket init
  2. Set up the PHP side (or examine the files in the php directory to integrate manually). For now, PHP resources aren't installed by PIP.

    yum install -y httpd php php-devel php-pear gcc
    pecl install channel://
    git clone git+
    cd perfbucket/php


  • Enable watching of a directory for xhprof/json files:

    perfbucket watch DIRECTORY
  • View the slowest requests by hour:

    perfbucket show <verbose> requests
  • View the slowest average pages by hour:

    perfbucket show <verbose> pages
  • Output the xhprof-compatible profiling data into a file:

    perfbucket export REQUEST-UUID > profile.xhprof_testing
  • Install the Cassandra schema:

    perfbucket init