May 22, Paul Wickman wrote forth saying... > - How often does gmond sample core metrics of the host it's running on?
take a look at ./gmond/metric.h .. each of the metrics are sampled/multicast using different time/value thresholds. the array at the bottom outlines it exactly. > - In /etc/gmond.conf, is "num_custom_metrics" a hard limit? What happens > if this is set to, say 10, and I report 12 metrics? this is not a hard-limit... it's just used to build the hash... if you set the size too small then the hash performance will drop. it's just a very rough estimate. > - In 'ganglia-rrd.pl', when an RRD database is updated for some host > metric, is it updated with the time "now" or the time listed in the > "REPORTED=" field from the XML output? in the old rrd backend.. it used the now() value.. gmetad uses the REPORTED field in the XML output though. > - Again in 'ganglia-rrd.pl', how does it behave when a host metric is > old? When updating an RRD with a time that has past, you'll get an error > saying as such. Will 'ganglia-rrd.pl' try to update (and thus get this > error from RRDs::update), or will it not even try? if a host is "dead" then the data doesn't even get written to the Round-Robin databases. -matt
