This question was recently posted to the discuss list, I am sending it
to the development list because it is related to a suggestion that Matt
posted several months ago, see the attached email.
Is this code to customize the rrd creation still going to make it into
the next or a future release of ganglia? Will the webfrontend scripts
read gmetad's config file so it will also know about the customized rrds
and what time windows are available to present the data or will similar
config options be in the webfrontend's conf.php?
I think this is an important feature that will unify the rrd creation.
Currently the step interval is defined by the data_source config option
while the rest of the rrd create options are hard coded into ganglia.
The original question below also brings up a problem that I have noticed
myself several times. The rrds are created with absolutely no room for
error when viewed by the webfrontend. The 15 second resolution rrds
contain exactly 60 minutes worth of data. Maybe large clusters create
some fuzzy timing issues with rrds being updated at the same time they
are being viewed. Whatever the reason, I have frequently seen rrd jump
to the day resolution when displaying data for the last hour.
The default rrd creation should probably contain a few extra data points
more than what will be used to display the data. It would also be nice
if the customized rrd option would be included in the next release.
~Jason
On Fri, 2004-12-10 at 05:16, Stephan Wonczak wrote:
> Hello, List!
>
> On Wed, 8 Dec 2004, Ron Reeder wrote:
>
> <snip>
>
> > The gmetad daemon was building the RRD database with 240 intervals of _13_
> > seconds, _not_ 15 seconds.
> > I don't know why gmetad was doing this ... I know it was from: rrdtool info
> > <hostname-metric.rrd>
> >
> > Since the number 240 for intervals is _hard coded_ in the gmetad server, I
> > don't know why you don't go
> > ahead and hard code the length of interval to 15 seconds... I have done
> > this at my site, and built
> > a patch and a new rpm based on that patch.
> <snip>
>
> Hmmm... This may be a FAQ or even a stupid question, but why is the
> number of intervals (an interval length) hardcoded at all? Isn't it
> possible to have this in the gmetad.conf?
> The reason for this question is that we would like to have some
> additional graphs: 6 hours and 10 minutes, both with a reasonable degree
> of resolution.
> Getting an 6-hour-graph is not too difficult; we managed to do this, but
> it uses the same data density as the daily-graph. We would like this to
> have the same resolution as the 'hourly'-graph (Ok, some data averaging is
> necessary here, too)
> Speaking of adding additional graph time intervals: It would be nice to
> have this a bit more accesible rather than buried inside the source of the
> web pages. Meaning: Inside gmetad.conf.
> Any comments are welcome, even those explaining why this is a stupid
> idea :-)
>
> Dipl. Chem. Dr. Stephan Wonczak
>
> Zentrum fuer Angewandte Informatik (ZAIK)
> Regionales Rechenzentrum der Universitaet zu Koeln (RRZK)
> Universitaet zu Koeln, Robert-Koch-Strasse 10, 50931 Koeln
> Tel: ++49/(0)221/478-5577, Fax: ++49/(0)221/478-5590
>
>
>
>
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from real users.
> Discover which products truly live up to the hype. Start reading now.
> http://productguide.itmanagersjournal.com/
> _______________________________________________
> Ganglia-general mailing list
> [EMAIL PROTECTED]
> https://lists.sourceforge.net/lists/listinfo/ganglia-general
--
/------------------------------------------------------------------\
| Jason A. Smith Email: [EMAIL PROTECTED] |
| Atlas Computing Facility, Bldg. 510M Phone: (631)344-4226 |
| Brookhaven National Lab, P.O. Box 5000 Fax: (631)344-7616 |
| Upton, NY 11973-5000 |
\------------------------------------------------------------------/
--- Begin Message ---
i just checked in changes to cvs and created a new 2.6.0 snapshot which
is available (along w/RPMs) at http://matt-massie.com/ganglia/
.
this version allows you to configure gmetad round-robin archives any way
that you like.
the default archives are (from the code)...
config->num_rras = 4;
config->rras = (char **) malloc (sizeof(char *) * config->num_rras);
/* 1 hour of 15 sec samples */
config->rras[0] = strdup("RRA:AVERAGE:0.5:15:240");
/* 1 day of 6 minute samples */
config->rras[1] = strdup("RRA:AVERAGE:0.5:360:240");
/* 1 month of hourly samples */
config->rras[2] = strdup("RRA:AVERAGE:0.5:3600:744");
/* 1 year of daily samples */
config->rras[3] = strdup("RRA:AVERAGE:0.5:86400:365");
people can alter the round-robin archive format via a gmetad.conf option
"round_robin_archives". it is a list of archives... for example..
round_robin_archives "RRA:AVERAGE:0.5:15:240"
would force gmetad to run with only one round-robin archive (1 hour of
15 second samples)
round_robin_archives "RRA:AVERAGE:0.5:1:3600" "RRA:AVERAGE:0.5:3600:240"
would force gmetad to run with two archives a (1 hour of 1 sec samples)
and (1 day of 6 minute samples)
to learn more about round-robin archives.. visit...
http://www.rrdtool.com/manual/rrdcreate.html
the data-source names, step and heartbeat are not configurable (and
really shouldn't be). altering them will prevent the front-end from
know where to find data... it needs to know the data-source DS names.
i haven't added the explicit NaN processing to gmetad... going to do
that now.
another quick question...
would anyone object if i changed to the location of the databases to use
the IP address of hosts instead of their fully-qualified name? there is
no nice answer here but i think it's the best of three evils.
feedback?
-matt
--
Mobius strippers never show you their back side
PGP fingerprint 'A7C2 3C2F 8445 AD3C 135E F40B 242A 5984 ACBC 91D3'
signature.asc
Description: This is a digitally signed message part
--- End Message ---
