alarm-threshold-create --query option is discarded
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
python-ceilometerclient |
Fix Released
|
High
|
Eoghan Glynn |
Bug Description
When creating an alarm with the new alarm-threshold
So for example using:
ceilometer alarm-threshold
--meter-name cpu_util --threshold 70.0 --comparison-
--period 600 --evaluation-
--alarm-action 'log://' \
--query resource_
results in a threshold alarm with a null query, whereas using the original alarm-create verb:
ceilometer alarm-create --name cpu_high --description 'instance running hot' \
--meter-name cpu_util --threshold 70.0 --comparison-
--period 600 --evaluation-
--alarm-action 'log://' \
--matching-
results in the constraint being used as expected.
This is a simple but nasty bug because the query is absolutely crucial to the evaluation of threshold alarms as it's used to constrain the periodic statistics query that determines the alarm state. Without these constraints for say an alarm on cpu_util for a particular instance would instead be applied to the cpu_util for *all* instances visible to the alarm owner.
Changed in python-ceilometerclient: | |
importance: | Undecided → High |
status: | New → In Progress |
assignee: | nobody → Eoghan Glynn (eglynn) |
Changed in python-ceilometerclient: | |
milestone: | none → 1.3.0 |
status: | Fix Committed → Fix Released |
Fix proposed to branch: master /review. openstack. org/68637
Review: https:/