[rrd-users] Re: Memory usage on large RRAs

Alex van den Bogaerdt alex at ergens.op.het.net
Sun Dec 4 23:16:11 MET 2005


On Sun, Dec 04, 2005 at 09:14:19PM +0100, Fabien Wernli wrote:

> I am monitoring ~700 nodes using rrdtool, each updating 7 databases on a
> central server.
> 
> Does anybody have experience on writing DEFs and CDEFs to minimize memory
> usage? At the time being I am unable to successfully plot a cumulative graph
> of all nodes, as memory runs out.

Each DEF and CDEF will consume memory.  You can start by getting rid of
as much CDEFs as you can.  Simple example:
  Don't do:
    DEF:ds0_bytes=....
    CDEF:ds0_bits=ds0_bytes,8,*
    CDEF:ds0_unknowntozero=ds0_bits,UN,0,ds0_bits,IF

  Do:
    DEF:ds0_bytes=....
    CDEF:ds0_unknowntozero=ds0_bytes,UN,0,ds0_bytes,8,*,IF


Another important step: make sure to use the right RRA.  If you are
going to graph many days in a single graph, it makes no sense to use
the 5-minute data to do so.  Using "now" (implied or not) for start
and/or end is almost always guarantied to cause this problem.  Make
sure to set start and end to a whole multiple of step*steps, and
make sure the amount of pixels corresponds with the amount of data.


Do you save lots of correlated stuff in the database?  If you are only
using two out of ten DSes, RRDtool still fetches all ten DSes.  It may
make sense to, for instance, keep ethernet errors in a different file.

-- 
Alex van den Bogaerdt
http://www.vandenbogaerdt.nl/rrdtool/

--
Unsubscribe mailto:rrd-users-request at list.ee.ethz.ch?subject=unsubscribe
Help        mailto:rrd-users-request at list.ee.ethz.ch?subject=help
Archive     http://lists.ee.ethz.ch/rrd-users
WebAdmin    http://lists.ee.ethz.ch/lsg2.cgi



More information about the rrd-users mailing list