Hi folks,

I modified rrdtool-1.2.12 so that I could specify a Holt-Winters
failure window-length > 28.

I've attached my patch that modifies the source in the two places where
the magic number 28 occurred (referred to as MAX_FAILURES_WINDOW_LEN).

It seems that since the beginning, the Holt-Winters code has had the
MAX_FAILURES_WINDOW_LEN set to 28.  Presumably, with 300 second step,
Jake found this to be sufficient since the window could be up to 2
hours 20 minutes, and he selected a default window-length of 9 (45
minutes).  However, I have some rrd files in which I've created with
step set to 1 second, so the window could only be 28 seconds in that
case.  (I update this RRD file periodically with a batch of new bit and
packet rates for each second.)

With my patch, I've had success setting window-length to 60 and
failure-threshold to 45 (seconds), but when I specify a window-length
that is fairly large (>60, such as 2700) I get this error:

   $ /usr/local/rrdtool-1.2.12/bin/rrdtool tune \
   one_sec_rates_dist_hw.rrd \
   --failure-threshold 2100 \
   --window-length 2700
   free(): invalid pointer 0x914e318!
   Memory fault
   $

gdb shows that the problem happens on this line of the function "rrd_free":

   if (rrd->rra_ptr) free(rrd->rra_ptr);

Any ideas why this happens?  Seems as though some pertinent data
structure is probably not be resized according to
MAX_FAILURES_WINDOW_LEN but it's not obvious to me why the rra_ptr is
being clobbered.

Dave

-- 
plonka@doit.wisc.edu  http://net.doit.wisc.edu/~plonka  ARS:N9HZF  Madison, WI


-- Attached file removed by Ecartis and put at URL below --
-- Type: text/plain
-- Size: 1k (1052 bytes)
-- URL : http://lists.ee.ethz.ch/p/rrdtool-window-length.diff


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