[rrd-users] Possible Bug?

[Chris Carpenter]

Hey Karl,

I'm not sure exactly what the symptom was that caused the segfault/crash but
the solution/cause was as you suggested, to reduce the number of rows. I had
assumed that the way rows worked was like in a spreadsheet/database where
each row is a specific instance of data. I'd assume that since it does
indeed work to store so much data without an increase in the number of rows,
then a row must mean something else in a rrd file?

In any case, I have tried doing the rrdtool graph command from outside cacti
and (with the old graph that had too many rows) it has the same issue, so it
doesn't appear to be anything directly to do with cacti.

Thanks for helping me fix my issue!
Chris Carpenter

P.S. I'm perfectly willing to help as much as I can if the developers do
want to look into fixing the crashing at the code level, even though it
isn't necessary to fix my particular problem.

On Thu, Mar 31, 2011 at 10:13 AM, Karl Fischer <rrd-users at ficos.de> wrote:

> Am 31.03.2011 15:57, schrieb Chris Carpenter:
> > Hello,
> >
> > I'm not sure whether your mailing list takes attachments,
> > so if you would just go to my cacti forum post
>
> Hi Chris,
>
> I just had a quick look at your post there and tried to reproduce the
> error on a freshly installed Debian squeeze but it works just fine
> out of the box using rrdtool_1.4.3-1 from the Debian distribution.
> (see graph attached)
>
> So this looks like you're hitting a race condition there.
> The segfault and the crash is definitely something for the
> rrd development team to look into.
>
> Knowing nothing about cacti I'm thinking it might run out of
> memory because you increased the number of rows so drastically.
>
> Try the same without changing the row counts (not necessary) ...
>
> - Karl
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/rrd-users/attachments/20110331/2d707d61/attachment.htm