Trouble with data input after migration to 64Bit CPU

[seveny]

Hi all,

I migrated my cacti-stuff from an old Athlon 1.3 GHz to a Athlon64 System (AMD Athlon(tm) 64 Processor 3000+). Both Suse Linux run with 2.6er kernels.
Had to use the rrdtool to manually migrate the rrd-files btw. Worked good.

SNMP-stuff works fine but my selfmade weatherstation script gets strange
timeouts. See for yourself:
http://www.knopfdruck.net/projekte/linuxwetter/

There are "holes" up to three croncycles (3x5 minutes) which I do not have
any explanation yet.

I turned on the DEBUG mode but my script gets parsed in these croncylces
just as in the others. Seems to be that the rrdtool update fails somehow? Is that possible?

Does anyone use a 64Bit CPU-System with selfmade data input methods?

--sven

[gandalf]

What is your Downed Host Detection set to? Running cacti 0.8.6h? Did you already change timeout values?
Reinhard

[seveny]

yes - Version 0.8.6h

host downed is "snmp reliable" but these scripts aren't connected to a host.
I changed it to "snmp and ping" for testing.

--sven

[gandalf]

... but these scripts aren't connected to a host.

I don't understand what you mean by that. I can't imagine a way to make cacti run some script without connecting this via Data Template/Graph Template to a host.
If that very host does not support SNMP, please leave the community string empty to suppress snmp checking. Verify Downed Host Detection using DEBUG mode. You'll find the results in log/cacti.log. If host is detected as down, cacti won't perform any script on that target
Reinhard

[seveny]

i have a weatherstation connected to the server directly via serial port.
i use "data input method" to collect data.
i create the graphs from the datasources which are filled by the data input method.
no host needed.
--sven
ps: in german: ich hab dazu mal einen kurzen artikel in der c't geschrieben (heft19/2005)