Chinaunix首页 | 论坛 | 博客
  • 博客访问: 5359666
  • 博文数量: 1144
  • 博客积分: 11974
  • 博客等级: 上将
  • 技术积分: 12312
  • 用 户 组: 普通用户
  • 注册时间: 2005-04-13 20:06
文章存档

2017年(2)

2016年(14)

2015年(10)

2014年(28)

2013年(23)

2012年(29)

2011年(53)

2010年(86)

2009年(83)

2008年(43)

2007年(153)

2006年(575)

2005年(45)

分类: LINUX

2010-09-11 21:26:15

#!/usr/bin/perl -w
#
# Acknowledgements:
#    - Sections of the code below were based on or, in some cases,
#    taken from examples provided in one or more of the following
#    places:
#      - Christiansen, Tom, and Nathan Torkington. Perl Cookbook.
#      1st ed. O'Reilly and Associates, 1998.
#      - Descartes, Alligator, and Tim Bunce. Programming the Perl
#      DBI. O'Reilly and Associates, 2000.
#      - Wall, Larry, Tom Christiansen, and Randal L. Schwartz.
#      Programming Perl. 2nd ed. O'Reilly and Associates, 1996.
#      - Szabo, Balazs. Documentation for
#      Parallel::ForkManager-0.7.5.
+/Parallel-ForkManager-0.7.5/ForkManager.pm . 03 Jan 2003.
#  Faults or incorrect implementations of code based upon code
#  from the above sources should not be attributed to original
#  source, but to my implementation.
#  - My thanks to those individuals visiting the Perl Monks
#  () "ChatterBox" between approximately
#  15:00 and 22:00 GMT on Friday, 03 Jan 2003, for answering
#  questions, especially the user known as Tye for his assistance
#  with questions while working out the child-to-parent messaging
#  methodology.
#  - If I have made an inappropriate attribution, or failed to
#  make an attribution, it was not intentional. Please contact me,
#  and I will be happy to attempt to correct the situation.
#
use DBI;
use Fcntl qw(:DEFAULT :flock);
use IO::Handle;
use Parallel::ForkManager;
use strict;
$| = 1;
my $table_name = "tablename";
my $cfg_data   = "data.file";
my $db_type  = "Oracle";
my $db_sid   = "databasename";
my $username = "username";
my $password = "password";
#
# Process reads information for a database to determine items to
#   handle, then handles each in a child process, as processing
#   times can be lengthy.
#
my $dbh_parent =
  DBI->connect( join ( ':', ( 'dbi', $db_type, $db_sid ) ),
    $username, $password, { AutoCommit => 0 } )
  or die (
    join (
        "\n",
        (
          'Database connection failed:',
        'Error returned was:',
        $DBI->errstr
        )
      )
      . "\n"
  );
$dbh_parent->{LongReadLen} = 64000;
my @targetlist = ();
{
    #
    # Get items to handle, and load into an array, so we do not need
    #   the parent db handle when we start creating child processes.
    #
    my ($item);
    my $sql_statement = "SELECT UNIQUE items FROM " . $tablename;
    my $statement_handle = $dbh_parent->prepare($sql_statement);
    $statement_handle->execute
      or die (
        join (
            "\n",
            (
              'Failure while executing SQL statement:',
            $sql_statement,
            'Error returned was:',
            $DBI->errstr
            )
          )
          . "\n"
      );
    my $result_handle =
      $statement_handle->bind_columns( \$item );
    #
    # Because child processes inherit the environment of their
    #   parent when spawned, they will also inherit the DB handles
    #   of the parent-thus, we get the listing of hosts into an
    #   array, then no longer need that handle (and prevent
    #   spurtious errors regarding the parent DB handle not being
    #   closed when child exits.
    #
    while ( $statement_handle->fetch ) {
        push ( @targetlist, $item );
    }
}
#
# Loop is done,
#   we no longer need the db connection from the parent process
#
$dbh_parent->disconnect;
#
# Code for returning a total count of modems handled at the end.
#   The pipe created should allow the children to send a value back
#   to the parent.
#
my $grandtotal = 0;
my $childcount = scalar(@targetlist);
pipe( READER, WRITER );
WRITER->autoflush(1);
my $MAX_PROCESSES = 7;
my $pm            = new Parallel::ForkManager($MAX_PROCESSES);
foreach my $item (@targetlist) {
    #
    # Fork off the next child process using Parallel::ForkManager to
    #   keep the number of children to no more than $MAX_PROCESSES.
    #
    $pm->start and next;
    # Close the read end of the pipe for the child processes.
    close(READER);
    #
    # Each child needs its own handle to the database,
    #   as they do not play well with each other's
    #
    my $dbh =
      DBI->connect( join ( ':', ( 'dbi', $db_type, $db_sid ) ),
        $username, $password, { AutoCommit => 0 } )
      or die (
        join (
            "\n",
            (
              'Database connection failed:',
            'Error returned was:',
            $DBI->errstr
            )
          )
          . "\n"
      );
    $dbh->{LongReadLen} = 64000;
    {
        my ($loop);
        #
        # Child processing occurs herein, including count of whatever
        #   is being done (stored in $loop).
        #
        #
        # ### CHILD PROCESS CODE HERE ####
        #
        #
        # Child processes in original script had to occasionally
        #   append to a file, thus the routine below for handling
        #   file locking.
        #  
        {
            #
            # Implemented using file locking to prevent data
            # corruption.
            #
            open( OUTFILE, ">>" . $cfg_data );
            {
                flock( OUTFILE, LOCK_EX );
                #
                # ### CODE TO ADD APPROPRIATE DATA TO FILE HERE ###
                #

            }
            close(OUTFILE);
        }
        #
        # Child process clean-up:
        #   - send item and count to parent via pipe
        #   - close DB handle
        #   - close writing end of the pipe for the child
        #   - exit child process
        #
        print( WRITER join ( ' ', ( $item, $loop ) ), "\n" );
        $dbh->disconnect;
    }
    close(WRITER);
    $pm->finish;
}
#
# Close the writing end of the pipe for the parent,
#   loop thru
#     read a number of lines equal to the number of children
#     expected, increment $grandtotal by the value read,
#   then close the read end of the pipe
#
# Data on a line sent from child to parent is in the form:
#         =
#         = [^\s]+
#   = \s+
#        =
#
{
    close(WRITER);
    for ( my $i = 0 ; $i < $childcount ; $i++ ) {
        my $line = ;
        chomp($line);
        my @parts = split ( /\s+/, $line, 2 );
        $grandtotal += $parts[1];
    }
    close(READER);
}
# Wait for any remaining children (should be none).
$pm->wait_all_children;
print <
Total items processed: $grandtotal
GRAND_TOTAL
 
阅读(1032) | 评论(0) | 转发(0) |
给主人留下些什么吧!~~