Here's a perl script that I use to just determine if my lines and routers
are up. You can easily modify it to suit your needs. A simple system
call would restart the webserver:
system("/etc/init.d/apache stop && /etc/init.d/apache start");
I run this in /etc/cron.d every minute and dump the results in a log upon
which I run logcheck.
I think it uses a default telnet timeout of 10 seconds. look at man
Net::Telnet to specify a timeout. It pings some hosts, and telnets to
port 80 on some webservers. Upon connecting to a webserver it issues,
"GET / HTTP/1.0" and expects the first line to be "HTTP/1.[01] 200 OK".
If not it prints an error. Note that some icmp is lost, being an
unreliable protocol. So for ping tests I try about 15 pings and if I
get at least 5 then I figure my line is up. Eventually I'd like to have
this dialup to an isp and email-page me if my line is really down.
Right now it just emails me.
Cory
line_monitor.pl
#!/usr/bin/perl -w
#Used to monitor t1 facilities as well as local webserver
use strict;
use Net::Ping;
use Net::Telnet;
sub ping ($);
sub port_test($$);
my $hostname = `hostname`; chomp($hostname);
my $progname = `basename $0`; chomp($progname);
my $date = `date +"%b %e %T"`; chomp($date);
my %ping_hosts = ( "123.456.789.00", "otherside_of_t1_router",
"123.456.789.00", "next_hop_router",
"64.58.76.227", "www.yahoo.com",
"128.223.142.13", "www.uoregon.edu"
);
# Use ping_hosts to determine if line is up
my $linetest=5;
for my $i (1..3) {
foreach(keys %ping_hosts) {
$linetest-- if &ping($_); # If can get 5 pings, line is probably up
}
}
if($linetest>0) {
system("traceroute www.yahoo.com -n -w 2 -m 10 2>&1");
# traceroute to yahoo
print "$date $hostname $progname: $ping_hosts{$linetest} ICMP Failed - Line
Down\n";
}
# Test webservers for connectivity
#my $retry = 0;
my %www_hosts = ( "192.168.0.20", "local_webserver",
"128.223.142.13", "www.uoregon.edu"
);
foreach(sort(keys %www_hosts)) {
my $r = &port_test($_, 80, "GET / HTTP/1.0\n\n");
unless ($r && $r=~m{HTTP/1.[01] 200 OK}) {
$r="" unless $r;
chomp($r);
print "$date $hostname $progname: $www_hosts{$_}:$_ WWW failed
$r\n";
}
}
exit(0);
#####################
# Functions
#####################
sub ping ($) {
my $host = $_[0]; # Host to ping
my $p = Net::Ping->new("icmp") or die "Can't create new ping object:
$!\n";
if ($p->ping ($host)) { return 1; }
else { return 0; }
$p->close();
}
sub port_test($$) {
my ($host, $port, $put) = @_;
my $t = new Net::Telnet(Host => $host, Port => $port,
Errmode => "return", Telnetmode => 0);
return 0 unless $t;
$t->put($put) if $put;
my $line = $t->getline();
$t->close();
return $line;
}
On Fri, Dec 13, 2002 at 12:12:20PM -0800, Tim Howe wrote:
> I have a heavily used web server that has been deciding to simply stop serving pages
>for up to 20 minutes at a time... I have already tried all manner of Apache and OS
>tweeks to stop this but nothing seems to work. Restarting the server puts everything
>back on track. What I would like to do, until I find a real fix, is have a Perl
>program try to connect to that machine on port 80, and if it fails to get a page
>within, say, 5 seconds, to restart the server.
> I have found a bunch of modules that will ping and or connect to web servers, but
>none of them seem to have a good way to time the response. Any suggestions?
>
> TimH
> _______________________________________________
> Eug-LUG mailing list
> [EMAIL PROTECTED]
> http://mailman.efn.org/cgi-bin/listinfo/eug-lug
>
_______________________________________________
Eug-LUG mailing list
[EMAIL PROTECTED]
http://mailman.efn.org/cgi-bin/listinfo/eug-lug
