masa16 / pwrake1

Obsolete verion of Pwrake: Parallel workflow extension for Rake

Pwrake

Parallel workflow extension for Rake

• Author: Masahiro Tanaka

(README in Japanese), (GitHub Repository)

Features

• Parallelize all tasks; no need to modify Rakefile, no need to use multitask.
• Tasks are executed in the given number of worker threads.
• Remote exuecution using SSH.
• Pwrake is an extension to Rake, not patch to Rake: Rake and Pwrake coexist.
• High parallel I/O performance using Gfarm file system.

Installation

Download source tgz/zip and expand, cd to subdir and install:

$ ruby setup.rb

Or, gem install:

$ gem install pwrake

Usage

Parallel execution using 4 cores at localhost:

$ pwrake -j 4

Parallel execution using all cores at localhost:

$ pwrake -j

Parallel execution using total 2*2 cores at remote 2 hosts:

1. Share your directory among remote hosts via distributed file system such as NFS, Gfarm.

2. Allow passphrase-less access via SSH in either way:

   • Add passphrase-less key generated by ssh-keygen. (Be careful)
   • Add passphrase using ssh-add.

3. Make hosts file in which remote host names and the number of cores are listed:

    $ cat hosts
    host1 2
    host2 2
   

4. Run pwrake with an option --hostfile or -F:

    $ pwrake --hostfile=hosts
   

Options

Command line option

    -F, --hostfile FILE              [Pw] Read hostnames from FILE
    -j, --jobs [N]                   [Pw] Number of threads at localhost (default: # of processors)
    -L, --logfile [FILE]             [Pw] Write log to FILE
        --ssh-opt, --ssh-option OPTION
                                     [Pw] Option passed to SSH
        --filesystem FILESYSTEM      [Pw] Specify FILESYSTEM (nfs|gfarm)
        --gfarm                      [Pw] FILESYSTEM=gfarm
    -A, --disable-affinity           [Pw] Turn OFF affinity (AFFINITY=off)
    -S, --disable-steal              [Pw] Turn OFF task steal
    -d, --debug                      [Pw] Output Debug messages
        --pwrake-conf [FILE]         [Pw] Pwrake configuation file in YAML
        --show-conf, --show-config   [Pw] Show Pwrake configuration options
        --report LOG                 [Pw] Report profile HTML from LOG and exit.

pwrake_conf.yaml

• If pwrake_conf.yaml exists at current directory, Pwrake reads options from it.

• Example (in YAML form):

    HOSTFILE : hosts
    LOGFILE : true
    TASKLOG : true
    PROFILE : true
    GNU_TIME : true
    PLOT_PARALLELISM : true
    DISABLE_AFFINITY: true
    DISABLE_STEAL: true
    FAILED_TARGET : delete
    PASS_ENV :
     - ENV1
     - ENV2
  

• Option list:

    HOSTFILE, HOSTS   default=false
    LOGFILE, LOG      default=none, string=filename, true="Pwrake%Y%m%d-%H%M%S_%$.log"
    TASKLOG           default=none, string=filename, true="Pwrake%Y%m%d-%H%M%S_%$.task"
    PROFILE           default=none, string=filename, true="Pwrake%Y%m%d-%H%M%S_%$.csv"
    WORK_DIR          default=$PWD
    FILESYSTEM        default=nil (autodetect)
    SSH_OPTION        (String) SSH option
    PASS_ENV          (Array) Environment variables passed to SSH
    GNU_TIME          If true, obtains PROFILEs using GNU time
    PLOT_PARALLELISM  If true, plot parallelism using GNUPLOT
    FAILED_TARGET     ( rename(default) | delete | leave ) failed files
    QUEUE_PRIORITY          RANK(default), FIFO, LIFO, LIHR
    NOACTION_QUEUE_PRIORITY FIFO(default), LIFO, RAND
    NUM_NOACTION_THREADS    default=4 when gfarm, else 1
    THREAD_CREATE_INTERVAL  default=0.01 (sec)
    HALT_QUEUE_WHILE_SEARCH true|false
    GRAPH_PARTITION         true|false
  

  for Gfarm system:

    DISABLE_AFFINITY  default=false
    DISABLE_STEAL     default=false
    STEAL_WAIT        default=0 (sec)
    STEAL_WAIT_MAX    default=10 (sec)
     : Wait min(STEAL_WAIT*2**n, STEAL_WAIT_MAX) sec for task steal.
    GFARM_BASEDIR     default="/tmp"
    GFARM_PREFIX      default="pwrake_$USER"
    GFARM_SUBDIR      default='/'
    MAX_GFWHERE_WORKER  default=8
  

Note for Gfarm

• gfwhere-pipe script (included in Pwrake) is used for file-affinity scheduling. This script requires Ruby/FFI (https://github.com/ffi/ffi). Install FFI by

    gem install ffi
  

For Graph Partitioning

• Compile and Install METIS 5.1.0 (http://www.cs.umn.edu/~metis/). This requires CMake.

• Install RbMetis (https://github.com/masa16/rbmetis) by

    gem install rbmetis -- \
     --with-metis-include=/usr/local/include \
     --with-metis-lib=/usr/local/lib
  

Tested Platform

• Ruby 2.1.4
• Rake 10.1.0
• CentOS 6.4

Acknowledgment

This work is supported by

• JST CREST, research area: "Development of System Software Technologies for Post-Peta Scale High Performance Computing," and
• MEXT Promotion of Research for Next Generation IT Infrastructure "Resources Linkage for e-Science (RENKEI)."