为什么我的 Gnocchi 测量的频率与设置的粒度不匹配

Question

我正在运行 openstack 并试图让我的 gnocchi 仪表更频繁地通过，以便我可以运行扩展演示而不会出现大量 5 分钟的滞后。 在 Gnocchi 中，我已将 Archive-policy 更改为粒度设置为 30 秒的自定义策略（我还使用现有的“中等”策略尝试了以下操作，结果相同）

+---------------------+--------------------------------------------------------+
| Field               | Value                                                  |
+---------------------+--------------------------------------------------------+
| aggregation_methods | std, count, min, max, sum, mean                        |
| back_window         | 0                                                      |
| definition          | - points: 120, granularity: 0:00:30, timespan: 1:00:00 |
| name                | test                                                    |
+---------------------+--------------------------------------------------------+

cpu_util 仪表正在获取正确性

+------------------------------------+-------------------------------------------------------------------+
| Field                              | Value                                                             |
+------------------------------------+-------------------------------------------------------------------+
| archive_policy/aggregation_methods | std, count, min, max, sum, mean                                   |
| archive_policy/back_window         | 0                                                                 |
| archive_policy/definition          | - points: 120, granularity: 0:00:30, timespan: 1:00:00            |
| archive_policy/name                | test                                                               |
| created_by_project_id              | e499d0c2e0fb4a05ac39c3f8c260052b                                  |
| created_by_user_id                 | 21759a51f3834b9bbae49c3ed17a13e4                                  |
| creator                            | 21759a51f3834b9bbae49c3ed17a13e4:e499d0c2e0fb4a05ac39c3f8c260052b |
| id                                 | e5a02f3a-9fbe-4e44-bb91-e1cfe6b86143                              |
| name                               | cpu_util                                                          |
| resource/created_by_project_id     | e499d0c2e0fb4a05ac39c3f8c260052b                                  |
| resource/created_by_user_id        | 21759a51f3834b9bbae49c3ed17a13e4                                  |
| resource/creator                   | 21759a51f3834b9bbae49c3ed17a13e4:e499d0c2e0fb4a05ac39c3f8c260052b |
| resource/ended_at                  | None                                                              |
| resource/id                        | 243b9715-95ba-4532-9728-3e61776e1c29                              |
| resource/original_resource_id      | 243b9715-95ba-4532-9728-3e61776e1c29                              |
| resource/project_id                | 43a7db62d5d54c4590e363868fff49e2                                  |
| resource/revision_end              | None                                                              |
| resource/revision_start            | 2018-08-08T14:05:09.770765+00:00                                  |
| resource/started_at                | 2018-08-08T13:20:45.948842+00:00                                  |
| resource/type                      | instance                                                          |
| resource/user_id                   | 4e5015006b304e7ca57edc5419b42be3                                  |
| unit                               | %                                                                 |
+------------------------------------+-------------------------------------------------------------------+

但测量结果仍然只每 5 分钟出现一次

gnocchi measures show e5a02f3a-9fbe-4e44-bb91-e1cfe6b86143

+---------------------------+-------------+--------------+
| timestamp                 | granularity |        value |
+---------------------------+-------------+--------------+
| 2018-08-08T13:30:00+00:00 |        30.0 | 0.0400002375 |
| 2018-08-08T13:35:00+00:00 |        30.0 | 0.0366666763 |
| 2018-08-08T13:40:00+00:00 |        30.0 | 0.0366667101 |
| 2018-08-08T13:45:00+00:00 |        30.0 | 0.0399999545 |
| 2018-08-08T13:50:00+00:00 |        30.0 | 0.0366664861 |
| 2018-08-08T13:55:00+00:00 |        30.0 | 0.0400000543 |
| 2018-08-08T14:00:00+00:00 |        30.0 | 0.0366665877 |
+---------------------------+-------------+--------------+

我有什么想法吗？

Answer 1

我遇到过同样的问题。 在 Gnocchi 支持的polling.yaml有一个新的配置文件： polling.yaml 。 正在那里设置资源轮询间隔。

---

• https://review.opendev.org/#/c/405682/
• https://docs.openstack.org/ceilometer/pike/admin/telemetry-best-practices.html