| 1 | # -*- perl -*- | ||
| 2 | |||
| 3 | # Copyright (c) 2009 AdCopy | ||
| 4 | # Author: Jeff Weisberg | ||
| 5 | # Created: 2009-May-12 11:03 (EDT) | ||
| 6 | # Function: monitor related processes | ||
| 7 | # | ||
| 8 | # $Id: Monitor.pm,v 1.17 2010/09/07 15:31:54 jaw Exp $ | ||
| 9 | |||
| 10 | # we periodically check the heartbeats of various processes (dancrs, scriblers, ...) | ||
| 11 | # and kibitz their info around the network | ||
| 12 | |||
| 13 | package AC::Yenta::Monitor; | ||
| 14 | use AC::Yenta::Debug 'monitor'; | ||
| 15 | use AC::Yenta::Config; | ||
| 16 | use AC::Yenta::Monitor::Client; | ||
| 17 | use AC::Misc; | ||
| 18 | use AC::Yenta::MySelf; | ||
| 19 | |||
| 20 | use Sys::Hostname; | ||
| 21 | use Socket; | ||
| 22 | use strict; | ||
| 23 | |||
| 24 | require 'AC/protobuf/yenta_status.pl'; | ||
| 25 | require 'AC/protobuf/heartbeat.pl'; | ||
| 26 | |||
| 27 | my $FREQ = 5; | ||
| 28 | my $OLD_DOWN = 30; | ||
| 29 | my $OLD_KEEP = 1800; | ||
| 30 | |||
| 31 | my %MON; # by 'id' (from config file) | ||
| 32 | |||
| 33 | sub init { | ||
| 34 | |||
| 35 | AC::DC::Sched->new( | ||
| 36 | info => 'monitor', | ||
| 37 | freq => $FREQ, | ||
| 38 | func => \&periodic, | ||
| 39 | ); | ||
| 40 | } | ||
| 41 | |||
| 42 | sub periodic { | ||
| 43 | |||
| 44 | my $mon = conf_value('monitor'); | ||
| 45 | |||
| 46 | # clean up old data | ||
| 47 | for my $id (keys %MON){ | ||
| 48 | isdown($id, 0) if $MON{$id}{lastup} < $^T - $OLD_DOWN; | ||
| 49 | } | ||
| 50 | |||
| 51 | # start monitoring (send heartbeat request) | ||
| 52 | for my $m (@$mon){ | ||
| 53 | my $ip = $m->{ipa}; | ||
| 54 | my $port = $m->{port}; | ||
| 55 | my $id = "$ip:$port"; | ||
| 56 | debug("start monitoring $id"); | ||
| 57 | |||
| 58 | my $ok = AC::Yenta::Monitor::Client->new( $ip, $port, | ||
| 59 | info => "monitor client: $id", | ||
| 60 | monitor_peer => $id, | ||
| 61 | ); | ||
| 62 | |||
| 63 | isdown($id, 0) unless $ok; | ||
| 64 | } | ||
| 65 | } | ||
| 66 | |||
| 67 | # data for kibitzing around | ||
| 68 | # array of ACPYentaStatus | ||
| 69 | sub export { | ||
| 70 | |||
| 71 | my @d; | ||
| 72 | my $here = my_datacenter(); | ||
| 73 | my $self = my_server_id(); | ||
| 74 | |||
| 75 | for my $v (values %MON){ | ||
| 76 | |||
| 77 | push @d, { | ||
| 78 | id => $v->{id}, # from config typically localhost:port | ||
| 79 | datacenter => $here, | ||
| 80 | via => $self, | ||
| 81 | hostname => $v->{hostname}, | ||
| 82 | subsystem => $v->{subsystem}, | ||
| 83 | environment => $v->{environment}, | ||
| 84 | status => $v->{status_code}, | ||
| 85 | timestamp => $v->{timestamp}, | ||
| 86 | lastup => $v->{lastup}, | ||
| 87 | sort_metric => $v->{sort_metric}, | ||
| 88 | server_id => $v->{server_id}, | ||
| 89 | instance_id => $v->{server_id}, | ||
| 90 | ip => $v->{ip}, | ||
| 91 | }; | ||
| 92 | } | ||
| 93 | return @d; | ||
| 94 | } | ||
| 95 | |||
| 96 | sub isdown { | ||
| 97 | my $id = shift; | ||
| 98 | my $code = shift; | ||
| 99 | |||
| 100 | my $d = $MON{$id}; | ||
| 101 | return unless $d; | ||
| 102 | |||
| 103 | # require 2 polls to fail | ||
| 104 | return unless $^T - $d->{lastup} >= 2 * $FREQ; | ||
| 105 | |||
| 106 | $code = 0 if $code == 200; | ||
| 107 | $d->{status_code} = $code || 0; | ||
| 108 | $d->{timestamp} = $^T; | ||
| 109 | |||
| 110 | debug("monitor $id is down"); | ||
| 111 | |||
| 112 | if( $d->{lastup} < $^T - $OLD_KEEP ){ | ||
| 113 | debug("monitor $id down too long. removing from report"); | ||
| 114 | delete $MON{$id}; | ||
| 115 | } | ||
| 116 | } | ||
| 117 | |||
| 118 | sub update { | ||
| 119 | my $id = shift; | ||
| 120 | my $gb = shift; # ACPHeartbeat | ||
| 121 | |||
| 122 | my $up; | ||
| 123 | eval { | ||
| 124 | $up = ACPHeartBeat->decode( $gb ); | ||
| 125 | $up->{id} = $id; | ||
| 126 | }; | ||
| 127 | if(my $e = $@){ | ||
| 128 | problem("cannot decode hb data: $e"); | ||
| 129 | return isdown($id, 0); | ||
| 130 | } | ||
| 131 | unless( $up->{status_code} == 200 ){ | ||
| 132 | return isdown($id, $up->{status_code}); | ||
| 133 | } | ||
| 134 | return isdown($id, 0) unless $^T - $up->{timestamp} < $OLD_DOWN; | ||
| 135 | |||
| 136 | debug("monitor $id is up"); | ||
| 137 | $up->{lastup} = $^T; | ||
| 138 | $up->{downcount} = 0; | ||
| 139 | |||
| 140 | _hb_ip_info( $up, $MON{$id} ); | ||
| 141 | $MON{$id} = $up; | ||
| 142 | } | ||
| 143 | |||
| 144 | sub _hb_ip_info { | ||
| 145 | my $up = shift; | ||
| 146 | my $old = shift; | ||
| 147 | |||
| 148 | my $ip; | ||
| 149 | |||
| 150 | $ip = $old->{ip} if ($old->{process_id} == $up->{process_id}) && ($old->{server_id} eq $up->{server_id}); | ||
| 151 | |||
| 152 | unless( $ip ){ | ||
| 153 | my $port = $up->{port}; | ||
| 154 | unless( $port ){ | ||
| 155 | # use monitored port (id is from config) | ||
| 156 | (undef, $port) = split /:/, $up->{id}; | ||
| 157 | } | ||
| 158 | |||
| 159 | if( $up->{ip} ){ | ||
| 160 | $ip = [ { ipv4 => $up->{ip}, port => $port, natdom => undef } ]; | ||
| 161 | }else{ | ||
| 162 | my $mynat = my_network_info(); | ||
| 163 | |||
| 164 | for my $i ( @$mynat ){ | ||
| 165 | push @$ip, { ipv4 => $i->{ipi}, port => $port, natdom => $i->{natdom} }; | ||
| 166 | } | ||
| 167 | } | ||
| 168 | } | ||
| 169 | |||
| 170 | $up->{ip} = $ip; | ||
| 171 | } | ||
| 172 | |||
| 173 | 1; |