NAME Net::FPing - quickly ping a large number of hosts SYNOPSIS use Net::FPing; DESCRIPTION This module was written for a single purpose only: sendinf ICMP EHCO REQUEST packets as quickly as possible to a large number of hosts (thousands to millions). It employs a sending thread and is fully event-driven (using AnyEvent), so you have to run an event model supported by AnyEvent to use this module. FUNCTIONS Net::FPing::ipv4_supported Returns true if IPv4 is supported in this module and on this system. Net::FPing::ipv6_supported Returns true if IPv6 is supported in this module and on this system. Net::FPing::icmp4_pktsize Returns the number of bytes each IPv4 ping packet has. Net::FPing::icmp6_pktsize Returns the number of bytes each IPv4 ping packet has. Net::FPing::icmp_ping [ranges...], $send_interval, $payload, \&callback Ping the given IPv4 address ranges. Each range is an arrayref of the form "[lo, hi, interval]", where "lo" and "hi" are octet strings with either 4 octets (for IPv4 addresses) or 16 octets (for IPV6 addresses), representing the lowest and highest address to ping (you can convert a dotted-quad IPv4 address to this format by using "inet_aton $address". The range "interval" is the minimum time in seconds between pings to the given range. If omitted, defaults to $send_interval. The $send_interval is the minimum interval between sending any two packets and is a way to make an overall rate limit. If omitted, pings will be send as fast as possible. The $payload is a 32 bit unsigned integer given as the ICMP ECHO REQUEST ident and sequence numbers (in unspecified order :). The request will be queued and all requests will be served by a background thread in order. When all ranges have been pinged, the "callback" will be called. Algorithm: Each range has an associated "next time to send packet" time. The algorithm loops as long as there are ranges with hosts to be pinged and always serves the range with the most urgent packet send time. It will at most send one packet every $send_interval seconds. This will ensure that pings to the same range are nicely interleaved with other ranges - this can help reduce per-subnet bandwidth while maintaining an overall high packet rate. The algorithm to send each packet is O(log n) on the number of ranges, so even a large number of ranges (many thousands) is managable. No storage is allocated per address. Performance: On my 2 GHz Opteron system with a pretty average nvidia gigabit network card I can ping around 60k to 200k adresses per second, depending on routing decisions. Example: ping 10.0.0.1-10.0.0.15 with at most 100 packets/s, and 11.0.0.1-11.0.255.255 with at most 1000 packets/s. Do not, however, exceed 1000 packets/s overall: my $done = AnyEvent->condvar; Net::FPing::icmp_ping [v10.0.0.1, v10.0.0.15, .01], [v11.0.0.1, v11.0.255.255, .001], .001, 0x12345678, sub { warn "all ranges pinged\n"; $done->broadcast; } ; $done->wait; Net::FPing::register_cb \&cb Register a callback that is called for every received ping reply (regardless of whether a ping is still in process or not and regardless of whether the reply is actually a reply ot a ping sent earlier). The code reference gets a single parameter - an arrayref with an entry for each received packet (replies are beign batched for greater efficiency). Each packet is represented by an arrayref with three members: the source address (an octet string of either 4 (IPv4) or 16 (IPv6) octets length), the payload as passed to "icmp_ping" and the round trip time in seconds. Example: a single ping reply with payload of 1 from "::1" gets passed like this: [ [ "\0\0\0\0\0\0\0\0\0\0\0\0\0\0\0\1", "0.000280141830444336", 1 ] ] Example: ping replies for 127.0.0.1 and 127.0.0.2, with a payload of 0x12345678: [ [ "\177\0\0\1", "0.00015711784362793", 305419896 ], [ "\177\0\0\2", "0.00090184211731", 305419896 ] ] Net::FPing::unregister_cb \&cb Unregister the callback again (make sure you pass the same codereference as to "register_cb"). AUTHOR Marc Lehmann http://home.schmorp.de/ AUTHOR This software is distributed under the GENERAL PUBLIC LICENSE, version 2 or any later.