On Mon, 2008-09-22 at 15:52 -0700, Bernard Li wrote: > Or you can re-post with a different topic here, and I'll try my best > to answer them :-)
We collect a half dozen or so simple metrics per node and add them to Ganglia using "gmetric". A naive implementation of this with 32 nodes each reporting approx eight metrics every thirty seconds achieved the desired result however cost of over 50% of our application performance. We have since changed the frequency from 30 seconds to five minutes and synchronised the daemons so they all run concurrently which has got us back most of the performance we lost. There is a specific requirement for this data to be monitored of I'd just turn the whole thing off. One problem with this that I'm still not happy with is we need to run a separate gmetric for each metric we collect, is there a way to pass gmetric a list of metrics so that it only has to startup and send data once rather than N times? Ashley. _______________________________________________ Beowulf mailing list, Beowulf@beowulf.org To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
