counting failures, success, and so on - giving wrong counts

[Klavs Klavsen]

I'm trying to let puppetdb query the result of a puppet run, and then fetch the numbers of failures, successes, changes etc.. using this:
curl -G 'http://localhost:8080/v3/aggregate-event-counts' --data-urlencode 'query=["=", "certname", "myhost.fqdn"]'  --data-urlencode 'summarize-by=containing-class'

This seems fine on some hosts, but on others it returns 'odd' values, like this:
{
  "successes" : 10,
  "failures" : 26,
  "noops" : 4,
  "skips" : 23,
  "total" : 36
}

Even though the puppet run says on the same machine says:

Info: Retrieving pluginfacts
Info: Retrieving plugin
Info: Loading facts
Info: Loading facts
Warning: The package type's allow_virtual parameter will be changing its default value from false to true in a future release. If you do not want to allow virtual packages, please explicitly set allow_virtual to false.
   (at /usr/lib/ruby/site_ruby/1.8/puppet/type/package.rb:430:in `default')
Info: Caching catalog for myhost.fqdn
Info: Applying configuration version '1431953336'
Notice: /Stage[main]/Selinux::Config/Exec[change-selinux-status-to-disabled]/returns: executed successfully
Notice: /Stage[main]/Logstash/Service[logstash]/ensure: ensure changed 'stopped' to 'running'
Info: /Stage[main]/Logstash/Service[logstash]: Unscheduling refresh on Service[logstash]
Notice: Finished catalog run in 9.96 seconds

How do I proceed, to figure out why those numbers are so warped ?

p.s. I'm using puppetdb-2.3.3 and puppet-3.7.4

[Klavs Klavsen]

I found a solution. adding --distinct (incl. start and end time, set to yesterday and today) - it works.