chore: address some gaps in k8s monitoring

[srikanthccv]

Summary

Part of #5373

1. Certain metrics like pod phase encode the state of the pod in value. This makes them non-aggregatable. Ideally, they should encode the state in labels. To address this, I added a special filters __value, which can be used to filter out metrics with value in the query. So if you want to count the number of pods in running state, you can __value=2. The same applies to node conditions. All the changes related to MetricValueFilter are doing this.

2. Change of SpaceAggregationSum to SpaceAggregationAvg. This is because it not meaningful to sum up the utilization percentage values. Example, if you have 2 pods with 1 cpu request and they are using 2 cpu each, then the utilization as a percentage is 200% for both. If you sum up the utilization, you get 400%, which is not correct (because the total cpu request is 2 and the total cpu usage is 4, which is 200% utilization).

3. Fix the map key for cpu_request and memory_request

4. Add pod phase and node condition metrics to the default metrics list.

---

Important

Enhances Kubernetes monitoring by adding value-based metric filtering, adjusting aggregation methods, and including new metrics for pod phases and node conditions.

• Behavior:
  • Introduces __value filter in query_builder.go to filter metrics by value, applicable in nodes_query.go, pods_query.go, and workload_query.go.
  • Changes SpaceAggregationSum to SpaceAggregationAvg in pods_query.go, workload_query.go, and query_builder.go to correctly calculate utilization percentages.
  • Adds pod phase and node condition metrics to default metrics in namespaces.go, nodes.go, and pods.go.
• Models:
  • Adds NodeCountByCondition and PodCountByPhase to infra.go for enhanced metric tracking.
• Misc:
  • Fixes map key for cpu_request and memory_request in workload_query.go.
  • Adds AddMetricValueFilter function in query_builder.go to support new filtering logic.

^{This description was created by for c795f58. It will automatically update as commits are pushed.}

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to c795f58 in 1 minute and 45 seconds

More details

• Looked at 782 lines of code in 16 files
• Skipped 0 files when reviewing.
• Skipped posting 11 drafted comments based on config settings.

1. pkg/query-service/app/metrics/query_builder.go:12

• Draft comment:
  The AddMetricValueFilter function has repetitive code for different numeric types. Consider using a type switch with a single case for all numeric types to reduce redundancy.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The PR introduces a new filter __value to handle non-aggregatable metrics by encoding state in labels. This is a good approach to make metrics aggregatable. However, the implementation of AddMetricValueFilter function in pkg/query-service/app/metrics/query_builder.go can be optimized. The function currently has repetitive code for different numeric types, which can be simplified.

2. pkg/query-service/app/inframetrics/pods_query.go:54

• Draft comment:
  The SpaceAggregation for utilization metrics should be SpaceAggregationAvg instead of SpaceAggregationSum. This change is necessary to correctly aggregate percentage values. Please ensure this change is applied consistently across all relevant queries.
• Reason this comment was not posted:
  Comment looked like it was already resolved.

3. pkg/query-service/app/metrics/v3/query_builder.go:336

• Draft comment:
  The AddMetricValueFilter function has repetitive code for different numeric types. Consider using a type switch with a single case for all numeric types to reduce redundancy.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The PR introduces a new filter __value to handle non-aggregatable metrics by encoding state in labels. This is a good approach to make metrics aggregatable. However, the implementation of AddMetricValueFilter function in pkg/query-service/app/metrics/query_builder.go can be optimized. The function currently has repetitive code for different numeric types, which can be simplified.

4. pkg/query-service/app/metrics/v4/query_builder.go:21

• Draft comment:
  The AddMetricValueFilter function has repetitive code for different numeric types. Consider using a type switch with a single case for all numeric types to reduce redundancy.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The PR introduces a new filter __value to handle non-aggregatable metrics by encoding state in labels. This is a good approach to make metrics aggregatable. However, the implementation of AddMetricValueFilter function in pkg/query-service/app/metrics/query_builder.go can be optimized. The function currently has repetitive code for different numeric types, which can be simplified.

5. pkg/query-service/app/inframetrics/namespaces.go:311

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Comment was on unchanged code.

6. pkg/query-service/app/inframetrics/nodes.go:329

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

7. pkg/query-service/app/inframetrics/pods.go:371

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

8. pkg/query-service/app/inframetrics/pods_query.go:203

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

9. pkg/query-service/app/metrics/query_builder.go:12

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

10. pkg/query-service/app/metrics/v3/query_builder.go:339

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

11. pkg/query-service/app/metrics/v4/cumulative/table.go:23

• Draft comment:
  Avoid using inline styles in React components. Instead, use external stylesheets, CSS classes, or styled components. This is applicable in multiple places in the code.
• Reason this comment was not posted:
  Marked as duplicate.

Workflow ID: wflow_09H8UZX5ndxmPLIu

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.