Nagios Support Forum

On my nagios dashboard, if I select graph by service, about half of the servers are missing the rrd file requested in the file list. THe file it is looking for looks like this command_name___%2Fhome%2Fdirectory_name.rrd. The folders storing the rrd files do have other rrd files for the same service and every graph for the host and service works fine. Only the one report, Graphs by Service is affected.

Consequently, the Graphs by Service for the plugin I wrote only has about half the graphs while others are broken.

There seems no apparent reason why this file is created for about 50% of the servers in the hostgroup but not the others. The installation is puppet controlled.

Could you show us a screenshot of what you're seeing as an example?

I can't without revealing some corporate info. The list for the graph by service report has mixed graph images and broken links to missing rrd files which are normally found in other server directories. Out of 40 servers, maybe 15 have graphs and the requested rrd file.

For service that doesn't have a graph, go to the Advanced tab.

In the "performance data" row, is there any data there?

markhorrocks wrote: the Graphs by Service for the plugin I wrote only has about half the graphs while others are broken.

missing rrd files are typically caused by the plugin not passing the perfdata correctly

check out the core 3 plugin api it describes how to correctly format the perfdata.

https://assets.nagios.com/downloads/nag ... inapi.html

Thanks @nozlaf!

@markhorrocks let us know if the previous posts help to solve this.

None of these answers seems to be the problem. The perfdata format is simple, 'label'=20 'label2'=30 'label3'=12 and so on. This works fine on about half the servers and all servers have good graphs when accessed by any means on the web dashboard except the GraphbyService report where half are good and half are links to missing files. Perhaps I should delete all rrd files for this service for the affected servers?

are you able to provide the actual perf output of an affected server and the perf output of a server which works? is it possible the ones which are not working are providing more than 4KB of perf data? (I have encountered this before but its a long shot I guess)

@markhorrocks, let us know if @nozlaf's suggestion works out for you.

In addition to that, follow the steps in this guide, even though it is for Nagios XI, the components should be the same and should shed some light on the problem.

Lets increase the logging verbosity and then take a deeper look into the logs. Follow the FAQ entry below to increase the log level of process_perfdata and npcd:

https://support.nagios.com/kb/article.php?id=9

Wait 15 - 20 minutes and then get a tail of the logs:
Send us a copy of /tmp/perfdata.txt and /tmp/npcd.txt

Don't forget to turn down the log level as per the FAQ when you are done!