On Sun, Sep 21, 2025 at 9:20 PM Nguyễn Hữu Khôi <nguyenhuukhoinw@gmail.com> wrote:

Hello,

I see that evaluation-periods and period option just works with "--type gnocchi_resources_threshold"

aodh alarm create --name cpu_hi --type gnocchi_resources_threshold --description 'instance running hot' --metric cpu_util --threshold 70.0 --comparison-operator gt --aggregation-method mean --period 60 --evaluation-periods 3 --alarm-action 'log://' --resource-id INSTANCE_ID --resource-type instance
+---------------------------+--------------------------------------+
| Field | Value |
+---------------------------+--------------------------------------+
| alarm_id | b30d83cd-2548-4964-a8fc-6c826bef5233 |
| name | cpu_hi |
| description | instance running hot |
| enabled | True |
| ok_actions | [] |
| alarm_actions | ['log:'] |
| insufficient_data_actions | [] |
| repeat_actions | False |
| type | gnocchi_resources_threshold |
| time_constraints | [] |
| project_id | 7fafdc0ed8f84cec9e3248966985533f |
| user_id | c9062ecef45646958894c50d95c02ce2 |
| timestamp | 2025-09-21T14:14:50.287176 |
| state | insufficient data |
| state_timestamp | 2025-09-21T14:14:50.287176 |
| state_reason | Not evaluated yet |
| severity | low |
| evaluate_timestamp | 2025-09-21T14:14:50.314318 |
| metric | cpu_util |
| resource_id | INSTANCE_ID |
| resource_type | instance |
| comparison_operator | gt |
| threshold | 70.0 |
| aggregation_method | mean |
| evaluation_periods | 3 |
| granularity | 60 |
+---------------------------+--------------------------------------+

It will be right if type prometheus can use two of these options too. Please correct me if it is our design for prometheus.

Nguyen Huu Khoi

On Sun, Sep 21, 2025 at 8:20 PM Nguyễn Hữu Khôi <nguyenhuukhoinw@gmail.com> wrote:
Hi,
I am using aodh latest and it is still not working for evaluation-periods and period option. Aodh still evaluates but it follows evaluation_interval in aodh.conf and just evaluator fist time then fire alarm,
Nguyen Huu Khoi

On Sun, Sep 21, 2025 at 8:05 PM Matthias Runge <mrunge@matthias-runge.de> wrote:
Hi there,

yes it does and it is also tested in CI.

Unfortunately I found the documentation for aodh missing
https://docs.openstack.org/aodh/latest/admin/telemetry-alarms.html

Reading your error message, I would think your version of aodh is not
new enough and much older than 2025.1, the code was added 2 years ago
https://opendev.org/openstack/aodh/commits/branch/master/aodh/evaluator/prometheus.py

Matthias

Am 19.09.25 um 16:49 schrieb Nguyễn Hữu Khôi:
> Hello,
>
> I would like to know if aodh supports evaluation-periods and period with
> prometheus query type?
>
> Any help would be appreciated.
>
> Thank you.
>
>
>
> Nguyen Huu Khoi
>
>
> On Fri, Sep 19, 2025 at 7:09 PM Nguyễn Hữu Khôi
> <nguyenhuukhoinw@gmail.com <mailto:nguyenhuukhoinw@gmail.com>> wrote:
>
> Hi,
>
> From https://opendev.org/openstack/python-observabilityclient/src/
> branch/master/observabilityclient/utils/metric_utils.py#L27
> <https://opendev.org/openstack/python-observabilityclient/src/
> branch/master/observabilityclient/utils/metric_utils.py#L27>
>
> We need nano /etc/openstack/prometheus.yaml with
>
> host: your-prometheus-host.example.com <http://your-prometheus-
> host.example.com>
> port: your prometheus port
>
> on where aodh_evaluator service is running.
>
> Nguyen Huu Khoi
>
>
> On Thu, Sep 18, 2025 at 9:25 PM Nguyễn Hữu Khôi
> <nguyenhuukhoinw@gmail.com <mailto:nguyenhuukhoinw@gmail.com>> wrote:
>
> Hello.
>
> I try to set up alarm such as:
>
> openstack alarm create --type prometheus --type prometheus --
> name memory_high_alarm --query
> 'memory_usage{resource_id="21d0792e-2d01-4df9-958a-
> d9018d13207f"}' --threshold 10
>
>
> I use ceilometer with prometheus backend but when it shows:
>
> +---------------------------
> +------------------------------------------------------------------+
> | Field | Value
> |
> +---------------------------
> +------------------------------------------------------------------+
> | alarm_id |
> cb7b8d07-0af1-4771-92c8-1a4c2ce4d1d1 |
> | name | memory_high_alarm
> |
> | description | prometheus alarm rule
> |
> | enabled | True
> |
> | ok_actions | []
> |
> | alarm_actions | []
> |
> | insufficient_data_actions | []
> |
> | repeat_actions | False
> |
> | type | prometheus
> |
> | time_constraints | []
> |
> | project_id | 31ef81adc8ad406e9e91e3eb8776c959
> |
> | user_id | 2e7a252719434d93ae0d270c8753294d
> |
> | timestamp | 2025-09-18T14:10:50.467568
> |
> | state | *insufficient data *
> |
> | state_timestamp | 2025-09-18T14:10:50.467568
> |
> | state_reason | Not evaluated yet
> |
> | severity | low
> |
> | evaluate_timestamp | 2025-09-18T14:10:50.485384
> |
> | comparison_operator | eq
> |
> | threshold | 10.0
> |
> | query |
> memory_usage{resource_id="21d0792e-2d01-4df9-958a-d9018d13207f"} |
> +---------------------------
> +------------------------------------------------------------------+
>
> and in aodh logs:
>
> 2025-09-18 21:20:02.442 1079 INFO aodh.evaluator [-] initiating
> evaluation cycle on 1 alarms
> 2025-09-18 21:20:02.442 1079 WARNING aodh.evaluator [-] Skipping
> alarm cb7b8d07-0af1-4771-92c8-1a4c2ce4d1d1, unsupported type:
> prometheus
>
> I am using openstack 2025.1
>
> I would like to get some guides on this, Many thanks.
>
> Nguyen Huu Khoi
>