I have always problems with capture rare problems with networks because simple programs such as ping or mtr doesn't logs problem in time scale.
I know that I could use SNMP and MRTG or similar tools but those are to big for small scale problems especially for particular server at client's network.
That's why I create simple script which logs timestamp when ICMP packets are dropped. Moreover it use standard tools such as ping and sh.
As usually use it on your own risk :)
#!/bin/sh
#
# Simple script for parallel ping