I would maybe try a different approach by accessing site information via TCP sockets? Using (use Socket) you cn verify an alive site by grabing some info from the page if you like... hell I have never tried this, but it should work right? Let me know if it does...

There is a wrapped up IO class called IO::Socket::INET that should for websites...

try

$client = IO::Socket::INET->new("PeerAddr => "the.host.com", PeerPort 
+=> 80, Type => SOCK_STREAM, Timeout => (How ever long you think it sh
+ould wait for a reply)) or die $@;
[download]

Then (in case you want to get some data from the connected socket)

bind(SOCKET, $client) or die "bind: $!";
[download]

if you get this far without error, then the site should be up and you can read from that socket and should be able to pick out bytes of the HTML markup... I think...

Read more on sockets to get ideas on how to read via $client->recv($data_read, $flags) or die...