.rrd update from cacti doesn't work as expected

[gromiT]

Hi,
I created several datasources with a data input of round trip time and packet loss. The graphs I created afterwards seemed to work fine, except that several graphs didn't show any values (nan)
I checked the .rrd files, all were present, but those where the graph didn't show any values weren't updated.
A check at the cacti log files shows the problem:

11/04/2002 5:15 PM - CMD: /usr/local/rrdtool-1.0.35/bin/rrdtool update /var/www/html/cacti/rra/device01.rrd --template out_ms:ploss N:1.749:0

11/04/2002 5:15 PM - CMD: /usr/local/rrdtool-1.0.35/bin/rrdtool update /var/www/html/cacti/rra/device02.rrd --template out_ms:out_ms:ploss:ploss N:0.00:4.533:0.01:0

So it's obvious that something went wrong when creating/updating that datasource.
Any ideas or suggestions ?

[raX]

Are both of those log entries using the same data input source? Because in the second entry I find this odd:

out_ms:out_ms:ploss:ploss N:0.00:4.533:0.01:0

Mainly because using the same field name twice in an output string won't work as expected.

NaN values imply that rrdtool is not getting valid data or any data at all. Even if your script is returning '0', that should still be reflected on the graph. What happens if you manually execute one of your log entries at the command line?

ie.

Code: Select all

/usr/local/rrdtool-1.0.35/bin/rrdtool update /var/www/html/cacti/rra/device01.rrd --template out_ms:ploss N:1.749:0

Tell me what you find.

-Ian

[gromiT]

Both log entries are using the same data input source (it's a perl script which pings a host n times; output is round trip time in msecs and packet loss in %)

As you pointed it out, it's weird that the same field is used twice (out_ms:out_ms:ploss:ploss...)
I forgot to point out that the update of device01 worked fine, only device02 didn't get any updates stored in the .rrd.
In fact my question was: Why does cacti use those same fields twice?

Actually, i work mainly with mozilla (1.1); after login to cacti, i had to reload the page to see anything at all. So is it possible that the problem is caused by the browser ? Right now, I added 7 more devices with the same data input source, now with IE 6, and everything seems to work fine

Anyway, cacti is a very fine piece of software !

[dbenders]

Currently I get this problems very frecuently, and my work around is to delete the RRA file, delete the DS from Cacti and start over.

I guess that there is a BUG somewhere, because this problems started with cacti 0.68a. In previows versions, it worked fine.

BTW: Can you please post your scripts here? I think that lot of guys will like to have it. Thanks!

[gromiT]

It seems to me that it is really a bug between Mozilla and Cacti; I added 10 more devices with IE6, and everything went ok.

If you're interested in the script, take a look at the following post:
http://www.raxnet.net/board/viewtopic.php?t=299
The script is actually from Brent Meshier, and you can find the graphs at his site:
http://www.meshier.com/cacti/graph_view.php?action=tree