[Ganglia restart extended metrics data lost] #240
Description
Linux CentOS 7, Ganglia 3.7.2 use python extended metrics.
When gmetad and gmond start at the same time, the metrics collects everything normally.
If gmond is turned off and gmond is restarted, the extended metrics will be lost with a high probability, with no new data or no RRD files. Restart gmond or gmetad repeatedly to generate data or RRD files again.
Network DNS /etc/hosts are properly configured , Ganglia log no ERROR, gmond UDP is sending data, guessing is gmetad's TCP problem.
What is the reason?