Poller not storing one script's data into RRD

[mfleetwo]

Hi,

I use a number of scripts gathering data and returning to Cacti, but after editing the Data Input Method and Data Template affecting one RRD it is not being updated.

I have debugged the script and it is being run and outputting the data in the format Cacti expects. I have also enabled Poller debugging but it doesn't report reading the data or doing rrdtool update. Poller is processing other data sources though.

How do I find out why poller.php isn't processing the data for this one Data Source?

Thanks,
Mike

Script debugging for non-working "sdlengineevents" and working "schedulequeue" Data Sources:

Code: Select all

# egrep '17:25.*(sdlengineevents|schedulequeue)' /var/log/ora_stats.log
2011-11-16T17:25:08.376 [15068] argv=['/usr/local/bin/ora_stats.py', '--log=/var/log/ora_stats.log', '--', 'cacti/c4ct1@LIVE', 'schedulequeuelen', 'schedulequeuetime']
2011-11-16T17:25:08.377 [15068] Printing statistics -> schedulequeuelen:0 schedulequeuetime:0
2011-11-16T17:25:28.597 [15354] argv=['/usr/local/bin/ora_stats.py', '--log=/var/log/ora_stats.log', '--group=database', '--', 'cacti/c4ct1@LIVE', 'newsdlengineevents', 'opensdlengineevents']
2011-11-16T17:25:29.143 [15354] Printing statistics -> newsdlengineevents:114 opensdlengineevents:48

Poller.php output only showing working "schedulequeue" Data Sources:

Code: Select all

# egrep '(sdlengineevents|schedulequeue)' /tmp/poller-20111116-1725.out 
11/16/2011 05:25:31 PM - POLLER: Poller[0] Parsed MULTI output field 'schedulequeuelen:0' [map schedulequeuelen->schedulequeuelen]
11/16/2011 05:25:31 PM - POLLER: Poller[0] Parsed MULTI output field 'schedulequeuetime:0' [map schedulequeuetime->schedulequeuetime]
11/16/2011 05:25:31 PM - POLLER: Poller[0] CACTI2RRD: /usr/local/rrdtool/bin/rrdtool update /srv/www/htdocs/cacti/rra/live_schedulequeue_2908.rrd --template schedulequeuelen:schedulequeuetime 1321464301:0:0

[noname]

Similar to these topics?

- http://forums.cacti.net/viewtopic.php?f=21&t=45249
- http://forums.cacti.net/viewtopic.php?f=21&t=45418
- http://forums.cacti.net/viewtopic.php?f=21&t=45447

[mfleetwo]

Finally got around to testing this further. Something strange is happening with cactid. Some times when testing cactid, it is not getting any output from my Data Source script:

Code: Select all

# ./cactid --readonly --stdout --verbosity=5
Host[0] DS[3229] WARNING: Result from SCRIPT not valid. Partial Result: ...
Host[0] DS[3229] SCRIPT: /usr/local/bin/ora_stats.py --log=/var/log/ora_stats.log --group=database -- cacti/c4ct1@LIVE newsdlengineevents opensdlengineevents, output: U

and at other times it seems to be working OK:

Code: Select all

# ./cactid --readonly --stdout --verbosity=5
Host[0] DS[3229] SCRIPT: /usr/local/bin/ora_stats.py --log=/var/log/ora_stats.log --group=database -- cacti/c4ct1@LIVE newsdlengineevents opensdlengineevents, output: newsdlengineevents:68 opensdlengineevents:37

but is seems to fail every time from cron. Also upgraded to the latest cactid-0.8.6k and got the same results.

Thanks,
Mike

[TheWitness]

Make sure that your script outputs all data at once. Spine and Cactid are both very temperamental, mainly for performance reasons. They don't wait around once you output data. So, doing the following is not supported in Cactid/Spine.

Code: Select all

echo "Name1:$Value1";
echo "Name2:$Value2\n";

The following is though:

Code: Select all

echo "Name1:$Value1 Name2:$Value2\n";

TheWitness

[mfleetwo]

My script is written in Python and does produce all its output with a single print statement, thus a single OS write() call.

[mfleetwo]

I have finally worked out the failure case. I have created 2 data sources the same, using 2 nearly identical data templates. The only difference is the working data template (and source) uses a script as the data input method where the single input parameter is set in the data source, but the failing data template (and source) uses a script as the data input method where three input parameters are set in the template and one is set in the data source.

In the failing case the RRD is never created

Manually created the RRD from the data source debug. Poller runs the script to fetch the data but never updates the DDR

[gandalf]

Please post screenshots of data template and data input method
R.

[mfleetwo]

Screen shots as requested.

• failing data template
  failing data input method
  working data input method
  working data template

[gandalf]

Please post both "Input String"s as a whole as only parts of them are shown.
R.

[mfleetwo]

Input strings, failing:
/usr/local/bin/ora_stats.py --log=/var/log/ora_stats.log --group=<group> -- <oracon> <statname1> <statname2>

Working:
/usr/local/bin/ora_stats.py --log=/var/log/ora_stats.log --group=database -- <oracon> opensdlengineevents newsdlengineevents

[gandalf]

And how do you provide values for <group> and stuff on the non-working data input method? This should be done when creating graphs, normally
R.

[mfleetwo]

For the non-working case, 3 of the values are set in the Data Template (see failing-data-template.png above) and the 4th value is set in the Data Source. Debugging in my script confirms cactid passes all the parameters to my script. Enabling debugging of the poller shows it never runs rrdtool update to put the values into the RRD.

(For the working case, the only value is set in the Data Source).

[gandalf]

Also upgraded to the latest cactid-0.8.6k and got the same results.

Sorry, that this occured quite late for me. You are indeed talking about "cactid 0.8.6"? This is years old and by no means supported any longer.
We now are at the end of 087, current version is 087i. We need this version on your side to reproduce any issues that still may be hidden in the code
R.

[mfleetwo]

I was talking about cactid (D), the the latest version still appears to be 0.8.6k (http://www.cacti.net/downloads/cactid/).

However we are using cacti (no-D) 0.8.6i. Will upgrade to 0.8.7i and try again.

[mfleetwo]

Now upgraded to cacti 0.8.7i with PIA 3.1.

As an aside, I was getting this in the output from poller.php. I had to move the definition of log_cacti_stats() above its use in poller.php to fix this error. Could be because I am using an old version of php, 4.3.8.

Code: Select all

Fatal error: Call to undefined function:  log_cacti_stats() in /srv/www/htdocs/cacti/poller.php on line 379

Still get the same results from the data sources. The one with parameters split between data template and data source still executes my script correctly but never updates the RRD file.