[rrd-developers] rrdcached + collectd issues
Thorsten von Eicken
tve at voneicken.com
Wed Oct 14 00:32:13 CEST 2009
Bernhard Reutner-Fischer wrote:
> On Tue, Oct 13, 2009 at 03:52:15AM -0700, Thorsten von Eicken wrote:
>> As I'm finishing to write this email rrdcached started to flush to disk.
>> So far nothing interesting happening (other than I/O). The connection
>> thread performance (or rather lack thereof) is virtually unchanged. I'll
>> grabs a fresh set of graphs as rrdtool-2*.png
> Do you, by chance, have figures for the just-rrdtool (with mmap)
> Just curious..
Bare collectd with rrdtool is instant death. Trying to do 5000 random
writes per second just doesn't fly on this hardware by an order of
magnitude. What I do have is collectd with rrdtool running on a tempfs
backed by the same disk storage. This works well but has pretty bad
behavior when the working set gets close to the memory available. Also,
if the set of RRDs changes over time (e.g. new servers replace old ones)
then you start getting into fragmentation issues. I can post some graphs
of what a server with this looks like later today.
More information about the rrd-developers