No announcement yet.

my ods (oreon 1.4) install experience

  • Filter
  • Time
  • Show
Clear All
new posts

  • my ods (oreon 1.4) install experience

    Maybe this can be of interest for people with the same configuration.
    Did all the things described here in different docs/posts (install procedure, ods database, patches, oreon/options/ods-options filled in etcetera)
    For perfparse I used the 'pipe to perfparsed' method, ODS doesn't work with pipes, so stopped perfparsed, removed the pipe-file service-perfdata.
    Because I compiled nagios with --with-default-perfdata I have no plugins or host_perfdata_command configured, Nagios writes the perfdata default.
    After a stop/start Nagios started to write the service-perfdata file.
    Started ODS, and indeed data was read in and displayed on the graphs.
    Very nice to have the same kind of graphs as with MRTG (day, week, month and year in one page)!
    ODS removes the service-perfdata file and my Nagios configuration doesn't seem to like that.
    I had to restart Nagios to make it write again, but because ODS is still running the service-perfdata gets removed right away so Nagios gets no change to write to the file..
    Too make a long story short, the only way I got it to work was by using a nagios-command to write a service-perfdata file.
    So I didn't use the libexec/process-service-perfdata plugin, which uses the old (nagios 1) macros by the way.
    Also this plugin doesn't do more than a normal command can do so I think a normal command is better/faster..
    Part of the configuration:
    define command{
    command_name process-service-perfdata
    command_line /usr/bin/printf "%b" "$LASTSERVICECHECK$\t$HOSTNAME$\t$SERVICEDESC$\t$S ERVICEOUTPUT$\t$SERVICESTATE$\t$SERVICEPERFDATA$\n " >> /usr/local/nagios/var/service-perfdata


    The service_perfdata_file setting has been made empty so this doesn't appear anymore in the nagios.cfg.

    By the way, the ods is now set up as a service, but there is in Nagios a possibility to configure a command to process the service-perfdata file in configurable intervals (service_perfdata_file_processing_command).
    Using that for the perfdata-part of ODS could have been possible too.
    Next thing I have to do is to look at the ODS purging routines, both for the logging and the perfdata.
    And these and logging lines are filling up my log too much, have to tune /etc/cron.d/oreon a bit.
    The last thing is that the graphs sometimes have gaps in them, maybe ODS isn't processing all the measurements.
    Have to look into those things, but so far, so good.

    Menno van Bennekom