State Meter for measuring of state changes and accurate uptime calculation.
A set of plugins similar to the instance meter where the gauge volume entry is used to represent current state (as based on nova vm_states, mapped to an int).
This should be a notification plugin, as well as a pollster, so that the meter is updated on an interval, but also on every state change in the VM.
This would provide an accurate set of state changes and allow for uptime calculations that take into account different states and are accurate down to the second.
Thus a user could calculated uptime for only given states (e.g. active, paused, building, etc), and ignore states such as suspended/stopped if they wanted.
Also be able to track state changes easily and graph/document them.
Implementation so far is avaible here:
https:/
I only have it working on Havana as master does not currently seem to have working notifications but can easily implement that once they are.
Base implementation is there, but work is needed in regards to testing, and also how to deal with vm_state -> int mappings. Have tried using nova.compute.
Example output:
+------
| Resource ID | Name | Type | Volume | Unit | Timestamp |
+------
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
| e19c3337-
+------
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Low
- Drafter:
- None
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- Obsolete
- Series goal:
- Accepted for future
- Implementation:
- Needs Code Review
- Milestone target:
- next
- Started by
- Adrian Turjak
- Completed by
- gordon chung
Related branches
Related bugs
Sprints
Whiteboard
Gerrit topic: https:/
Addressed by: https:/
Adds a state metric to ceilometer
can someone confirm that this is covered by the following patch: https:/
Has indeed been covered by that patch, feel free to close this.
-Adrian Turjak