File Coverage

lib/SNMP/Effective.pm

Criterion	Covered	Total	%
statement	10	12	83.3
branch	1	2	50.0
condition			n/a
subroutine	4	4	100.0
pod			n/a
total	15	18	83.3

package SNMP::Effective;

=head1 NAME

SNMP::Effective - An effective SNMP-information-gathering module

=head1 VERSION

1.1101

=head1 SYNOPSIS

    use SNMP::Effective;

    my $snmp = SNMP::Effective->new(
        max_sessions => $NUM_POLLERS,
        master_timeout => $TIMEOUT_SECONDS,
    );

    $snmp->add(
        dest_host => $ip,
        callback => sub { store_data() },
        get => [ '1.3.6.1.2.1.1.3.0', 'sysDescr' ],
    );

    # lather, rinse, repeat

    # retrieve data from all hosts
    $snmp->execute;

=head1 DESCRIPTION

This module collects information, over SNMP, from many hosts and many OIDs,
really fast.

It is a wrapper around the facilities of C, which is the Perl
interface to the C libraries in the C package. Advantages of using
this module include:

=over 4

=item Simple configuration

The data structures required by C are complex to set up before
polling, and parse for results afterwards. This module provides a simpler
interface to that configuration by accepting just a list of SNMP OIDs or leaf
names.

=item Parallel execution

Many users are not aware that C can poll devices asynchronously
using a callback system. By specifying your callback routine as in the
L section above, many network devices can be polled in parallel,
making operations far quicker. Note that this does not use threads.

=item It's fast

To give one example, C can walk, say, eight indexed OIDs
(port status, errors, traffic, etc) for around 300 devices (that's 8500 ports)
in under 30 seconds. Storage of that data might take an additional 10 seconds
(depending on whether it's to RAM or disk). This makes polling/monitoring your
network every five minutes (or less) no problem at all.

=back

The interface to this module is simple, with few options. The sections below
detail everything you need to know.

=head1 METHODS ARGUMENTS

The method arguments are very flexible. Any of the below acts as the same:

    $obj->method(MyKey => $value);
    $obj->method(my_key => $value);
    $obj->method(My_Key => $value);
    $obj->method(mYK__EY => $value);

=cut

use warnings;
use strict;
use constant DEBUG => $ENV{'SNMP_EFFECTIVE_DEBUG'} ? 1 : 0;
use SNMP;
use SNMP::Effective::Host;
use SNMP::Effective::HostList;
use Time::HiRes qw/usleep/;
use POSIX qw(:errno_h);

use base qw/ SNMP::Effective::Dispatch /;

our $VERSION = '1.1101';
our %SNMPARG = (
    Version => '2c',
    Community => 'public',
    Timeout => 1e6,
    Retries => 2
);

=head1 ATTRIBUTES

=head2 master_timeout

Get/Set the master timeout

=head2 max_sessions

Get/Set the number of max session

=head2 hostlist

Returns a list containing all the hosts.

=head2 arg

Returns a hash with the default args

=head2 callback

Returns a ref to the default callback sub-routine.

=head2 heap

Returns a value for the default heap.

=cut

BEGIN {
    no strict 'refs';
    my %sub2key = qw/
                      max_sessions    maxsessions
                      master_timeout  mastertimeout
                      _varlist        _varlist
                      hostlist        _hostlist
                      arg             _arg
                      callback        _callback
                      heap            _heap
                  /;

    for my $subname (keys %sub2key) {
        *$subname = sub {
            my($self, $set) = @_;
            $self->{ $sub2key{$subname} } = $set if(defined $set);
            $self->{ $sub2key{$subname} };
        }
    }
}

=head1 METHODS

=head2 new

This is the object constructor, and returns a L object.

=head3 Arguments

=over 4

=item C

Maximum number of simultaneous SNMP sessions.

=item C

Maximum number of seconds before killing execute.

=back

All other arguments are passed on to $snmp_effective->add( ... ).

=cut

sub new {
    my $class = shift;
    my %args = _format_arguments(@_);
    my $self = (ref $class) ? $class : $class->_new_object(%args);

    $self->add(%args);

    return $self;
}

sub _new_object {
    my $class = shift;
    my %args = @_;

    return bless {
        maxsessions => 1,
        mastertimeout => undef,
        _sessions => 0,
        _hostlist => SNMP::Effective::HostList->new,
        _varlist => [],
        _arg => {},
        _callback => sub {},
        %args,
    }, $class;
}

=head2 C

Adding information about what SNMP data to get and where to get it.

=head3 Arguments

=over 4

=item dest_host

Either a single host, or an array-ref that holds a list of hosts. The format
is whatever L can handle.

=item C

A hash-ref of options, passed on to SNMP::Session.

=item C

A reference to a sub which is called after each time a request is finished.

=item C

This can hold anything you want. By default it's an empty hash-ref.

=item C / C / C

Either "oid object", "numeric oid", L or an
array-ref containing any combination of the above.

=item C

Either a single L or a L or an array-ref of any of
the above.

=back

This can be called with many different combinations, such as:

=over 4

=item C / any other argument

This will make changes per dest_host specified. You can use this to change arg,
callback or add OIDs on a per-host basis.

=item C / C / C / C

The OID list submitted to L will be added to all dest_host, if no
dest_host is specified.

=item C / C

This can be used to alter all hosts' SNMP arguments or callback method.

=back

=cut

sub add {
    my $self = shift;
    my %in = _format_arguments(@_) or return;
    my $hostlist = $self->hostlist;
    my $varlist = $self->_varlist;
    my @new_varlist;

    # setup desthost input argument
    if($in{'desthost'} and ref $in{'desthost'} ne 'ARRAY') {
        $in{'desthost'} = [$in{'desthost'}];
        warn "Adding host(@{ $in{'desthost'} })" if DEBUG;
    }

    # add to varlist
    for my $key (keys %SNMP::Effective::Dispatch::METHOD) {
        next unless($in{$key});
        $in{$key} = [$in{$key}] unless(ref $in{$key});

        if(@{$in{$key}}) {
            warn "Adding $key(@{ $in{$key} })" if DEBUG;
            unshift @{$in{$key}}, $key;
            push @new_varlist, $in{$key};
        }
    }

    $in{'arg'} ||= delete $in{'args'};

    if(ref $in{'desthost'} eq 'ARRAY') {
        for my $addr (@{$in{'desthost'}}) {

            # add/update hosts
            my $host = $hostlist->get_host($addr)
                    || $hostlist->add_host(
                           address => $addr,
                           arg => $in{'arg'} || $self->arg,
                           callback => $in{'callback'} || $self->callback,
                           heap => $in{'heap'} || $self->heap,
                       );

            push @$host, (@$host or @new_varlist) ? @new_varlist : @$varlist;
            $host->arg($in{'arg'}) if($in{'arg'});
            $host->callback($in{'callback'}) if($in{'callback'});
            $host->heap($in{'heap'}) if($in{'heap'});
        }
    }
    else {

        # update $self with generic args
        $self->arg($in{'arg'}) if(ref $in{'arg'} eq 'HASH');
        $self->callback($in{'callback'}) if(ref $in{'callback'});
        $self->heap($in{'heap'}) if(exists $in{'heap'});

        # update $self and all hosts with @new_varlist
        if(@new_varlist) {
            push @$varlist, @new_varlist;
            for my $host (values %$hostlist) {
                push @$host, @new_varlist;
            }
        }
    }

    return 1;
}

=head2 execute

This method starts setting and/or getting data. It will run as long as
necessary, or until L seconds has passed. Every time some
data is set and/or retrieved, it will call the callback-method, as defined
globally or per host.

=cut

sub execute {
    my $self = shift;

    unless(scalar($self->hostlist)) {
        return 0;
    }

    $self->_init_lock;

    if(my $timeout = $self->master_timeout) { # dispatch with master timeout
        my $die_msg = "alarm_clock_timeout";

        warn "Execute dispatcher with timeout ($timeout)" if DEBUG;

        eval {
            local $SIG{'ALRM'} = sub { die $die_msg };
            alarm $timeout;
            $self->dispatch and SNMP::MainLoop();
            alarm 0;
        };

        # check for timeout
        if($@ and $@ =~ /$die_msg/mx) {
            $self->master_timeout(0);
            warn "Master timeout!" if DEBUG;
            SNMP::finish();
        }
        elsif($@) {
            die $@;
        }
    }
    else { # dispatch without master timeout
        warn "Execute dispatcher without timeout" if DEBUG;
        $self->dispatch and SNMP::MainLoop();
    }

    return 1;
}

sub _create_session {
    local $! = 0;

    my($self, $host) = @_;
    my $snmp = SNMP::Session->new(%SNMPARG, $host->arg);

    unless($snmp) {
        my($retry, $msg) = $self->_check_errno($!);
        warn "SNMP session failed for host $host: $msg" if DEBUG;
        return $retry ? '' : undef;
    }

    warn "SNMP session created for $host" if DEBUG;

    return $snmp;
}

sub _check_errno {
    my($self, $err) = @_;
    my $retry = 0;
    my $errstr = '';

    if(not $err) {
        $errstr = "Couldn't resolve hostname";
    }
    elsif($errstr = "$err") {
        if(
            $err == EINTR  || # Interrupted system call
            $err == EAGAIN || # Resource temp. unavailable
            $err == ENOMEM || # No memory (temporary)
            $err == ENFILE || # Out of file descriptors
            $err == EMFILE    # Too many open fd's
        ) {
            $errstr .= ' (will retry)';
            $retry = 1;
        }
    }

    return $retry, $errstr;
}

=head1 FUNCTIONS

=head2 C

Takes two arguments: One OID to match against, and the OID to match.

    match_oid("1.3.6.10",   "1.3.6");    # return 10
    match_oid("1.3.6.10.1", "1.3.6");    # return 10.1
    match_oid("1.3.6.10",   "1.3.6.11"); # return undef

=cut

sub match_oid {
    my $p = shift or return;
    my $c = shift or return;
    return ($p =~ /^ \.? $c \.? (.*)/mx) ? $1 : undef;
}

=head2 C

Inverse of make_numeric_oid: Takes a list of mib-object strings, and turns
them into numeric format.

 make_numeric_oid("sysDescr"); # return .1.3.6.1.2.1.1.1

=cut

sub make_numeric_oid {
    my @input = @_;

    for my $i (@input) {
        next if($i =~ /^ [\d\.]+ $/mx);
        $i = SNMP::translateObj($i);
    }

    return wantarray ? @input : $input[0];
}

=head2 C

Takes a list of numeric OIDs and turns them into an mib-object string.

    make_name_oid("1.3.6.1.2.1.1.1"); # return sysDescr

=cut

sub make_name_oid {
    my @input = @_;

    for my $i (@input) {
        $i = SNMP::translateObj($i) if($i =~ /^ [\d\.]+ $/mx);
    }

    return wantarray ? @input : $input[0];

}

sub _format_arguments {
    return if(@_ % 2 == 1);

    my %args = @_;

    for my $k (keys %args) {
        my $v = delete $args{$k};
        $k = lc $k;
        $k =~ s/_//gmx;
        $args{$k} = $v;
    }

    return %args;
}

sub _init_lock {
    my $self = shift;

    pipe my $READ, my $WRITE or die "Failed to create pipe: $!";
    select +( select($READ), $| = 1 )[0];
    select +( select($WRITE), $| = 1 )[0];
    print $WRITE "\n";

    warn "Lock is ready and unlocked" if DEBUG;

    return $self->{'_lock_fh'} = [ $READ, $WRITE ];
}

sub _wait_for_lock {
    my $self = shift;
    my $LOCK_FH = $self->{'_lock_fh'}->[0];

    warn "Waiting for lock to unlock..." if DEBUG;
    defined readline $LOCK_FH or die "Failed to read from LOCK_FH: $!";
    warn "The lock is now locked again" if DEBUG;

    return 1;
}

sub _unlock {
    my $self = shift;
    my $LOCK_FH = $self->{'_lock_fh'}->[1];

    warn "Unlocking lock" if DEBUG;
    print $LOCK_FH "\n";

    return 1;
}

=head1 THE CALLBACK METHOD

When C is done collecting data from a host, it calls a callback
method, provided by the C<< Callback => sub{} >> argument. Here is an example of a
callback method:

 sub my_callback {
     my($host, $error) = @_

     if($error) {
         warn "$host failed with this error: $error"
         return;
     }

     my $data = $host->data;

     for my $oid (keys %$data) {
         print "$host returned oid $oid with this data:\n";

         print join "\n\t",
               map { "$_ => $data->{$oid}{$_}" }
                   keys %{ $data->{$oid}{$_} };
         print "\n";
     }
 }

=head1 DEBUGGING

Debugging is enabled through setting the environment variable

    SNMP_EFFECTIVE_DEBUG=1 perl myscript.pl

It will print the debug information to STDERR.

=head1 NOTES

=over 4

=item C

L doesn't really do a SNMP native "walk". It makes a series
of "getnext", which is almost the same as SNMP's walk.

=item C

If you want to use SNMP SET, you have to build your own varbind:

 $varbind = SNMP::VarBind($oid, $iid, $value, $type);
 $effective->add( set => $varbind );

=back

=head1 AUTHOR

Jan Henning Thorsen, C<<  >>

=head1 BUGS

Please report any bugs or feature requests to
C, or through the web interface at
L.
I will be notified, and then you'll automatically be notified of progress on
your bug as I make changes.

=head1 ACKNOWLEDGEMENTS

Various contributions by Oliver Gorwits.

Sigurd Weisteen Larsen contributed with a better locking mechanism.

=head1 COPYRIGHT & LICENSE

Copyright 2007 Jan Henning Thorsen, all rights reserved.

This program is free software; you can redistribute it and/or modify it
under the same terms as Perl itself.

=cut

1;


1;