|
Packit |
90a5c9 |
#!@perlbin@
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# Licensed to the Apache Software Foundation (ASF) under one or more
|
|
Packit |
90a5c9 |
# contributor license agreements. See the NOTICE file distributed with
|
|
Packit |
90a5c9 |
# this work for additional information regarding copyright ownership.
|
|
Packit |
90a5c9 |
# The ASF licenses this file to You under the Apache License, Version 2.0
|
|
Packit |
90a5c9 |
# (the "License"); you may not use this file except in compliance with
|
|
Packit |
90a5c9 |
# the License. You may obtain a copy of the License at
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# http://www.apache.org/licenses/LICENSE-2.0
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# Unless required by applicable law or agreed to in writing, software
|
|
Packit |
90a5c9 |
# distributed under the License is distributed on an "AS IS" BASIS,
|
|
Packit |
90a5c9 |
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
|
Packit |
90a5c9 |
# See the License for the specific language governing permissions and
|
|
Packit |
90a5c9 |
# limitations under the License.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# logresolve.pl
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# v 1.2 by robh imdb.com
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# usage: logresolve.pl <infile >outfile
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# input = Apache/NCSA/.. logfile with IP numbers at start of lines
|
|
Packit |
90a5c9 |
# output = same logfile with IP addresses resolved to hostnames where
|
|
Packit |
90a5c9 |
# name lookups succeeded.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# this differs from the C based 'logresolve' in that this script
|
|
Packit |
90a5c9 |
# spawns a number ($CHILDREN) of subprocesses to resolve addresses
|
|
Packit |
90a5c9 |
# concurrently and sets a short timeout ($TIMEOUT) for each lookup in
|
|
Packit |
90a5c9 |
# order to keep things moving quickly.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# the parent process handles caching of IP->hostnames using a Perl hash
|
|
Packit |
90a5c9 |
# it also avoids sending the same IP to multiple child processes to be
|
|
Packit |
90a5c9 |
# resolved multiple times concurrently.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# Depending on the settings of $CHILDREN and $TIMEOUT you should see
|
|
Packit |
90a5c9 |
# significant reductions in the overall time taken to resolve your
|
|
Packit |
90a5c9 |
# logfiles. With $CHILDREN=40 and $TIMEOUT=5 I've seen 200,000 - 300,000
|
|
Packit |
90a5c9 |
# logfile lines processed per hour compared to ~45,000 per hour
|
|
Packit |
90a5c9 |
# with 'logresolve'.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# I haven't yet seen any noticeable reduction in the percentage of IPs
|
|
Packit |
90a5c9 |
# that fail to get resolved. Your mileage will no doubt vary. 5s is long
|
|
Packit |
90a5c9 |
# enough to wait IMO.
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# Known to work with FreeBSD 2.2
|
|
Packit |
90a5c9 |
# Known to have problems with Solaris
|
|
Packit |
90a5c9 |
#
|
|
Packit |
90a5c9 |
# 980417 - use 'sockaddr_un' for bind/connect to make the script work
|
|
Packit |
90a5c9 |
# with linux. Fix from Luuk de Boer <luuk_de_boer pi.net>
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
require 5.004;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
$|=1;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
use FileHandle;
|
|
Packit |
90a5c9 |
use Socket;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
use strict;
|
|
Packit |
90a5c9 |
no strict 'refs';
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
use vars qw($PROTOCOL);
|
|
Packit |
90a5c9 |
$PROTOCOL = 0;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my $CHILDREN = 40;
|
|
Packit |
90a5c9 |
my $TIMEOUT = 5;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my $filename;
|
|
Packit |
90a5c9 |
my %hash = ();
|
|
Packit |
90a5c9 |
my $parent = $$;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my @children = ();
|
|
Packit |
90a5c9 |
for (my $child = 1; $child <=$CHILDREN; $child++) {
|
|
Packit |
90a5c9 |
my $f = fork();
|
|
Packit |
90a5c9 |
if (!$f) {
|
|
Packit |
90a5c9 |
$filename = "./.socket.$parent.$child";
|
|
Packit |
90a5c9 |
if (-e $filename) { unlink($filename) || warn "$filename .. $!\n";}
|
|
Packit |
90a5c9 |
&child($child);
|
|
Packit |
90a5c9 |
exit(0);
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
push(@children, $f);
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
&parent;
|
|
Packit |
90a5c9 |
&cleanup;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
## remove all temporary files before shutting down
|
|
Packit |
90a5c9 |
sub cleanup {
|
|
Packit |
90a5c9 |
# die kiddies, die
|
|
Packit |
90a5c9 |
kill(15, @children);
|
|
Packit |
90a5c9 |
for (my $child = 1; $child <=$CHILDREN; $child++) {
|
|
Packit |
90a5c9 |
if (-e "./.socket.$parent.$child") {
|
|
Packit |
90a5c9 |
unlink("./.socket.$parent.$child")
|
|
Packit |
90a5c9 |
|| warn ".socket.$parent.$child $!";
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
sub parent {
|
|
Packit |
90a5c9 |
# Trap some possible signals to trigger temp file cleanup
|
|
Packit |
90a5c9 |
$SIG{'KILL'} = $SIG{'INT'} = $SIG{'PIPE'} = \&cleanup;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my %CHILDSOCK;
|
|
Packit |
90a5c9 |
my $filename;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
## fork child processes. Each child will create a socket connection
|
|
Packit |
90a5c9 |
## to this parent and use an unique temp filename to do so.
|
|
Packit |
90a5c9 |
for (my $child = 1; $child <=$CHILDREN; $child++) {
|
|
Packit |
90a5c9 |
$CHILDSOCK{$child}= FileHandle->new;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
if (!socket($CHILDSOCK{$child}, AF_UNIX, SOCK_STREAM, $PROTOCOL)) {
|
|
Packit |
90a5c9 |
warn "parent socket to child failed $!";
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
$filename = "./.socket.$parent.$child";
|
|
Packit |
90a5c9 |
my $response;
|
|
Packit |
90a5c9 |
do {
|
|
Packit |
90a5c9 |
$response = connect($CHILDSOCK{$child}, sockaddr_un($filename));
|
|
Packit |
90a5c9 |
if ($response != 1) {
|
|
Packit |
90a5c9 |
sleep(1);
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
} while ($response != 1);
|
|
Packit |
90a5c9 |
$CHILDSOCK{$child}->autoflush;
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
## All child processes should now be ready or at worst warming up
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my (@buffer, $child, $ip, $rest, $hostname, $response);
|
|
Packit |
90a5c9 |
## read the logfile lines from STDIN
|
|
Packit |
90a5c9 |
while(<STDIN>) {
|
|
Packit |
90a5c9 |
@buffer = (); # empty the logfile line buffer array.
|
|
Packit |
90a5c9 |
$child = 1; # children are numbered 1..N, start with #1
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# while we have a child to talk to and data to give it..
|
|
Packit |
90a5c9 |
do {
|
|
Packit |
90a5c9 |
push(@buffer, $_); # buffer the line
|
|
Packit |
90a5c9 |
($ip, $rest) = split(/ /, $_, 2); # separate IP form rest
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
unless ($hash{$ip}) { # resolve if unseen IP
|
|
Packit |
90a5c9 |
$CHILDSOCK{$child}->print("$ip\n"); # pass IP to next child
|
|
Packit |
90a5c9 |
$hash{$ip} = $ip; # don't look it up again.
|
|
Packit |
90a5c9 |
$child++;
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
} while (($child < ($CHILDREN-1)) and ($_ = <STDIN>));
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
## now poll each child for a response
|
|
Packit |
90a5c9 |
while (--$child > 0) {
|
|
Packit |
90a5c9 |
$response = $CHILDSOCK{$child}->getline;
|
|
Packit |
90a5c9 |
chomp($response);
|
|
Packit |
90a5c9 |
# child sends us back both the IP and HOSTNAME, no need for us
|
|
Packit |
90a5c9 |
# to remember what child received any given IP, and no worries
|
|
Packit |
90a5c9 |
# what order we talk to the children
|
|
Packit |
90a5c9 |
($ip, $hostname) = split(/\|/, $response, 2);
|
|
Packit |
90a5c9 |
$hash{$ip} = $hostname;
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# resolve all the logfiles lines held in the log buffer array..
|
|
Packit |
90a5c9 |
for (my $line = 0; $line <=$#buffer; $line++) {
|
|
Packit |
90a5c9 |
# get next buffered line
|
|
Packit |
90a5c9 |
($ip, $rest) = split(/ /, $buffer[$line], 2);
|
|
Packit |
90a5c9 |
# separate IP from rest and replace with cached hostname
|
|
Packit |
90a5c9 |
printf STDOUT ("%s %s", $hash{$ip}, $rest);
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
########################################
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
sub child {
|
|
Packit |
90a5c9 |
# arg = numeric ID - how the parent refers to me
|
|
Packit |
90a5c9 |
my $me = shift;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# add trap for alarm signals.
|
|
Packit |
90a5c9 |
$SIG{'ALRM'} = sub { die "alarmed"; };
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# create a socket to communicate with parent
|
|
Packit |
90a5c9 |
socket(INBOUND, AF_UNIX, SOCK_STREAM, $PROTOCOL)
|
|
Packit |
90a5c9 |
|| die "Error with Socket: !$\n";
|
|
Packit |
90a5c9 |
$filename = "./.socket.$parent.$me";
|
|
Packit |
90a5c9 |
bind(INBOUND, sockaddr_un($filename))
|
|
Packit |
90a5c9 |
|| die "Error Binding $filename: $!\n";
|
|
Packit |
90a5c9 |
listen(INBOUND, 5) || die "Error Listening: $!\n";
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
my ($ip, $send_back);
|
|
Packit |
90a5c9 |
my $talk = FileHandle->new;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# accept a connection from the parent process. We only ever have
|
|
Packit |
90a5c9 |
# have one connection where we exchange 1 line of info with the
|
|
Packit |
90a5c9 |
# parent.. 1 line in (IP address), 1 line out (IP + hostname).
|
|
Packit |
90a5c9 |
accept($talk, INBOUND) || die "Error Accepting: $!\n";
|
|
Packit |
90a5c9 |
# disable I/O buffering just in case
|
|
Packit |
90a5c9 |
$talk->autoflush;
|
|
Packit |
90a5c9 |
# while the parent keeps sending data, we keep responding..
|
|
Packit |
90a5c9 |
while(($ip = $talk->getline)) {
|
|
Packit |
90a5c9 |
chomp($ip);
|
|
Packit |
90a5c9 |
# resolve the IP if time permits and send back what we found..
|
|
Packit |
90a5c9 |
$send_back = sprintf("%s|%s", $ip, &nslookup($ip));
|
|
Packit |
90a5c9 |
$talk->print($send_back."\n");
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# perform a time restricted hostname lookup.
|
|
Packit |
90a5c9 |
sub nslookup {
|
|
Packit |
90a5c9 |
# get the IP as an arg
|
|
Packit |
90a5c9 |
my $ip = shift;
|
|
Packit |
90a5c9 |
my $hostname = undef;
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# do the hostname lookup inside an eval. The eval will use the
|
|
Packit |
90a5c9 |
# already configured SIGnal handler and drop out of the {} block
|
|
Packit |
90a5c9 |
# regardless of whether the alarm occurred or not.
|
|
Packit |
90a5c9 |
eval {
|
|
Packit |
90a5c9 |
alarm($TIMEOUT);
|
|
Packit |
90a5c9 |
$hostname = gethostbyaddr(gethostbyname($ip), AF_INET);
|
|
Packit |
90a5c9 |
alarm(0);
|
|
Packit |
90a5c9 |
};
|
|
Packit |
90a5c9 |
if ($@ =~ /alarm/) {
|
|
Packit |
90a5c9 |
# useful for debugging perhaps..
|
|
Packit |
90a5c9 |
# print "alarming, isn't it? ($ip)";
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
# return the hostname or the IP address itself if there is no hostname
|
|
Packit |
90a5c9 |
$hostname ne "" ? $hostname : $ip;
|
|
Packit |
90a5c9 |
}
|
|
Packit |
90a5c9 |
|
|
Packit |
90a5c9 |
|