#!/usr/bin/perl

my $greeting = $ARGV[0] // '********** PERL RECEIVED A NOTIFICATION:';

sub my_receiver {
    print $greeting, "\n";
}

NetSNMP::TrapReceiver::register("all", \&my_receiver) || warn "failed 
+to register\n";
print STDERR "Loaded the example perl snmptrapd handler\n";
[download]

perl do "/usr/local/share/snmp/mytrapd.pl hello";
[download]

This modification doesn't work (no output is produced and as far as I can tell, my handler is not being called). In my real world scenario, the command line argument happens to be a configuration file that contains extensive configuration information that I don't want hard coded into the code. I could of course hard code the path to the configuration file which I have done as a work around but I am wondering if anyone knows how to make this DWIM?

Update: I have received the following response from the module author which may mean what I am trying to accomplish is impossible unless kill -HUP <pid> causes the snmptrapd daemon to re-read the configuration file and re-register the function.

That won't work.  The perl 'do' command only expects a filename.

Instead, you can either define a subroutine in the file rather than ha
+ve
the file itself do something.  IE, in the file if you put:

sub foo {
   print "$_[0]\n";
}

and then put these lines in the snmptrapd.conf file:

perl do /path/to/script
perl foo("hello world");
perl foo("now I am passing something different");
[download]

Update 2: The author has told me that kill -HUP <pid> doesn't do what I want so the best I can hope for is hard coding the path to the configuration file in the script and then having the function periodically reload it.