Add support to gather per device metrics for instances
The existing measurements only allow us to gather disk.* data at the aggregate level. As in if an instance has multiple disk's associated to them, you get the cumulative read/write requests and rates. This doesnt help if you want to bill the user on each volume.
We need to be able to gather per volume measurements for each instance along with cumulative.
The approach here is to leverage existing polling we have to capture the cumulative disk metrics and
isolate the per disk read/write requests and bytes.
The per disk metrics will be their own individual meters with dynamic disk name as part of the metric names. Following are the meter names for each as part of the setup.cfg:
'disk.device.
'disk.device.
'disk.device.
'disk.device.
'disk.device.
'disk.device.
'disk.device.
'disk.device.
Each of the above metrics will be part of their own pollster which is derived from the cumulative data. The resource metadata associate with each sample will have the device name. For example,
{'device' : <device_name>}
Since this blueprint is a holdover from icehouse, we don't have a separate spec in ceilometer-specs repo.
Blueprint information
- Status:
- Complete
- Approver:
- Eoghan Glynn
- Priority:
- Medium
- Drafter:
- Pradeep Kilambi
- Direction:
- Approved
- Assignee:
- Pradeep Kilambi
- Definition:
- Approved
- Series goal:
- Accepted for juno
- Implementation:
- Implemented
- Milestone target:
- 2014.2
- Started by
- Pradeep Kilambi
- Completed by
- Pradeep Kilambi
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Support to preserve per disk metrics
You should not set a milestone target unless the blueprint has been properly prioritized by the project drivers.
Addressed by: https:/
Support for per disk volume measurements