Closed shtripat closed 7 years ago
@anmolbabu can you have a look at the https://github.com/rochaporto/collectd-ceph/tree/master/plugins This plugin as well uses json output and pushes the details to time series DB. I understand that in case of json we would have to calculate the percentage usage value but with suggestion from ceph team on https://bugzilla.redhat.com/show_bug.cgi?id=1360230, isnt it doable.
Probably, this needs to be either of the below:
The pool object in tendrl ceph integration definitions, will provide a way to indicate a way to read the >utilization data from some directory in etcd if monitoring is enabled and the tendrl-api will look into this path >to provide the utilization data or The performance-monitoring application will periodically update the utilization data to a location in etcd >and also the performance monitoring application will expose apis to access this data and the tendrl-api will >proxy to it.
I personally think that approach 2 although is a 2-step process, is safer and better than approach 1 of >involving changes to definitions on the fly once the monitoring is enabled.
Will change the spec as suggested
@anmolbabu https://github.com/Tendrl/specifications/issues/44 talks about this tag based job execution and its linking with role based job execution on node agents.
Signed-off-by: Shubhendu shtripat@redhat.com