settingsLogin | Registersettings

[Openstack-operators] [telemetry] Issues with autoscaling using HEAT, Aodh and Ceilometer

0 votes

Hi

I have installed OpenStack using OpenStack-Ansible 14.1.1. I have installed
Ceilometer, Aodh and using Mongo DB. I am trying simple autoscaling
scenario. On Horizon GUI, I have not been able to see the resource usage
getting updated consistently. In "ceilometer meter-list" command is also
not showing the update list of resources. In compute node ceilometer
polling, I am seeing log like "Skip pollster cpu_util, no resources found
this cycle"
. Could you give any direction that what could be the issue?

Only couple of times, I have seen alarm state being changed to "alarm" from
"insufficient data" but at this point Aodh Notifier gives error:
http://paste.openstack.org/show/612823/
http://paste.openstack.org/show/612823/
.

Please give some pointers to find the actual root cause behind the issues.

Regards,
Amit


OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
asked Jun 18, 2017 in openstack-operators by Amit_Kumar (980 points)   2 5 8

1 Response

0 votes

On 16/06/17 04:32 AM, Amit Kumar wrote:
Hi

I have installed OpenStack using OpenStack-Ansible 14.1.1. I have
installed Ceilometer, Aodh and using Mongo DB. I am trying simple
autoscaling scenario. On Horizon GUI, I have not been able to see the
resource usage getting updated consistently. In "ceilometer meter-list"
command is also not showing the update list of resources. In compute
node ceilometer polling, I am seeing log like /"Skip pollster cpu_util,
no resources found this cycle"/. Could you give any direction that what
could be the issue?

is ansible 14.x mapped to newton? just an fyi, ceilometer's legacy
storage functionality (mongodb) is deprecated as of Ocata. also, the
horizon resource usage is similarly deprecated (if not removed). we
advise you to use gnocchi+grafana[1] for storage (or another db).

regarding Skip pollster cpu_util log, that is normal (specifically if
you're using libvirt/kvm). it's just logging that because libvirt
doesn't provide that metric via polling so it's calculated in post.

[1] http://gnocchi.xyz

--
gord


OpenStack-operators mailing list
OpenStack-operators@lists.openstack.org
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
responded Jun 18, 2017 by gordon_chung (19,300 points)   2 3 8
...