Hi, Burr -

On Tue, May 03, 2016 at 10:48:23PM +0530, Burr Sutter wrote:

> Any concerns of the use of this metric capture/display tool within the CDK?
> https://docs.openshift.com/enterprise/latest/install_config/cluster_metrics.html

We may be aiming a little low.  Heapster/Hawkular/K8S integration is
compelling, but Heapster is not that rich a source of data, compared
to something like Performance Co-Pilot [2].  Compare dozens of metrics
[1] from the former (specialized to containers) to the thousands from
the latter (containers -and- host) [3].

Luckily, PCP is packageable as a container, and can already talk [4]
to Hawkular via the latter's graphite protocol adapter [5], to
supplement the Heapster data, so it can be just a
deployment-configuration problem, not a "need more code" one.

[1] https://github.com/kubernetes/heapster/blob/master/docs/storage-schema.md
[2] http://www.pcp.io/
[3] http://paste.fedoraproject.org/362338/17510146 (a random subset; 
/usr/bin/pminfo)
[4] http://man7.org/linux/man-pages/man1/pcp2graphite.1.html
[5] https://github.com/hawkular/hawkular-metrics/tree/master/clients/ptranslator

- FChE

_______________________________________________
Devtools mailing list
[email protected]
https://www.redhat.com/mailman/listinfo/devtools

Reply via email to