Context: -------- I teach people the basics of telemetry, alarms and autoscaling. I run Ceilometer with a 60 sec polling interval and want to see measures and trigger alarms after 2-3 minutes. My environment is a DevStack cloned from the master branch some two weeks ago. Ceilometer uses Gnocchi as publisher. nova-manage version returns 17 (Queens). Problem: ------- It takes 10 minutes or longer (I have seen over 30 minutes) after instance launch until the instance's measures are posted to Gnocchi. The notification agent log then includes the message "Resource <UUID> created". Before that, there is no trace of that UUID in the log.
Next problem is a lag in publishing measurements: $ date Wed Nov 22 07:16:38 UTC 2017 $ gnocchi measures show --resource-id 4ec5c876-367a-4918-9037-671b5ece3665 cpu_util +---------------------------+-------------+----------------+ | timestamp | granularity | value | +---------------------------+-------------+----------------+ | 2017-11-22T06:00:00+00:00 | 3600.0 | 0.128224536267 | | 2017-11-22T06:48:00+00:00 | 60.0 | 0.1357511344 | ... | 2017-11-22T06:56:00+00:00 | 60.0 | 0.1310314404 | +---------------------------+-------------+----------------+ If you compare the timestamp of the last measurement with the current time, there is a difference of 20 minutes. As a result, my alarms are in status "insufficient data", since Aodh doesn't find measurements for the last few minutes. Is this a bug, is this normal? And mostly - can this behaviour be changed, i.e. can I tell Ceilometer to post the measures to Gnocchi faster and can I remove this lag? Thanks much, Bernd Bausch _______________________________________________ Mailing list: http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack Post to : openstack@lists.openstack.org Unsubscribe : http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
