[PR-3] Runtime Integration

[khansaad]

Description

This PR integrates multi-datasource layer handling into the runtime recommendation flow so detected layer information is available during recommendation processing.

Fixes # (issue)

Type of change

• Bug fix
• New feature
• Docs update
• Breaking change (What changes might users need to make in their application due to this PR?)
• Requires DB changes

How has this been tested?

Please describe the tests that were run to verify your changes and steps to reproduce. Please specify any test configuration required.

• New Test X
• Functional testsuite

Test Configuration

• Kubernetes clusters tested on:

Checklist 🎯

• Followed coding guidelines
• Comments added
• Dependent changes merged
• Documentation updated
• Tests added or updated

Additional information

Include any additional information such as links, test results, screenshots here

Summary by Sourcery

Add multi-datasource support to experiments and integrate it into layer detection, validation, and recommendation flows while preserving backward compatibility with the existing single datasource field.

New Features:

• Support configuring multiple datasources per experiment via new datasources fields in API and internal model.
• Enable runtime recommendation and metrics collection flows to work with multi-datasource setups, requiring a Prometheus datasource for PromQL-based queries.

Bug Fixes:

• Prevent attempts to run runtime recommendations or metrics collection when no serviceable datasource is configured, emitting clear validation and runtime warnings instead.

Enhancements:

• Extend layer detection to use configured datasources and surface detected layer information into recommendation processing.
• Improve datasource validation to check all configured datasource names and provide clearer error messages when none are serviceable.
• Adjust accelerator detection logic to skip Prometheus-specific checks when only non-Prometheus datasources such as Cryostat are available.

[sourcery-ai]

Reviewer's Guide

Integrates multi-datasource awareness (with Prometheus as mandatory for metrics) throughout experiment validation, layer detection, and recommendation runtime, while preserving backward compatibility with the legacy single datasource field and adding Cryostat-specific handling for metrics and accelerator detection paths.

Sequence diagram for Prometheus datasource selection in RecommendationEngine

sequenceDiagram
    participant RecommendationEngine
    participant KruizeObject
    participant DataSourceCollection
    participant DataSourceInfo

    RecommendationEngine->>KruizeObject: getDatasources()
    alt datasources not empty
        loop for each ds in datasources
            RecommendationEngine->>DataSourceCollection: getInstance()
            RecommendationEngine->>DataSourceCollection: getDataSourcesCollection()
            RecommendationEngine->>DataSourceCollection: get(ds)
            DataSourceCollection-->>RecommendationEngine: DataSourceInfo
            alt DataSourceInfo not null
                RecommendationEngine->>DataSourceInfo: getProvider()
                alt provider is PROMETHEUS
                    RecommendationEngine->>RecommendationEngine: set dataSource = ds
                    RecommendationEngine->>RecommendationEngine: break
                end
            end
        end
    else datasources empty
        RecommendationEngine->>KruizeObject: getDataSource()
        alt dataSource not null
            RecommendationEngine->>DataSourceCollection: getInstance()
            RecommendationEngine->>DataSourceCollection: getDataSourcesCollection()
            RecommendationEngine->>DataSourceCollection: get(dataSource)
            DataSourceCollection-->>RecommendationEngine: DataSourceInfo
            alt DataSourceInfo null or provider not PROMETHEUS
                RecommendationEngine->>RecommendationEngine: set dataSource = null
            end
        end
    end

    alt dataSource is null
        RecommendationEngine->>RecommendationEngine: throw Exception
    end

Loading

Sequence diagram for PromQL datasource resolution with Cryostat

sequenceDiagram
    participant RecommendationEngine
    participant DataSourceCollection
    participant DataSourceInfo

    RecommendationEngine->>RecommendationEngine: fetchMetricsBasedOnProfileAndDatasource(..., dataSourceInfo, ...)
    alt [dataSourceInfo.provider is CRYOSTAT]
        RecommendationEngine->>RecommendationEngine: promQLDataSourceInfo = dataSourceInfo
        RecommendationEngine->>DataSourceCollection: getInstance()
        RecommendationEngine->>DataSourceCollection: getDataSourcesCollection()
        loop for each ds in values()
            RecommendationEngine->>DataSourceInfo: getProvider()
            alt [provider is PROMETHEUS]
                RecommendationEngine->>RecommendationEngine: promQLDataSourceInfo = ds
                RecommendationEngine->>RecommendationEngine: loop break
            end
        end
    else [dataSourceInfo.provider is not CRYOSTAT]
        RecommendationEngine->>RecommendationEngine: promQLDataSourceInfo = dataSourceInfo
    end

    RecommendationEngine->>RecommendationEngine: fetchContainerMetricsBasedOnDataSourceAndProfile(..., promQLDataSourceInfo, ...)

    alt [promQLDataSourceInfo.provider is PROMETHEUS]
        RecommendationEngine->>RecommendationEngine: accelerator detection enabled
    else [non-PROMETHEUS]
        RecommendationEngine->>RecommendationEngine: skip accelerator detection
    end

Loading

File-Level Changes

Change	Details	Files
Add multi-datasource support to experiment and API models with backward compatibility for the legacy single datasource field.	Deprecate single datasource field but retain it for backward compatibility in KruizeObject and CreateExperimentAPIObject. Introduce datasources list field on both experiment domain and API objects with getters/setters. Implement getDatasources() helpers that adapt a legacy single datasource into a list or return an empty list when unset. Wire datasources from CreateExperimentAPIObject into KruizeObject in the Converters utility.	src/main/java/com/autotune/analyzer/kruizeObject/KruizeObject.java src/main/java/com/autotune/analyzer/serviceObjects/CreateExperimentAPIObject.java src/main/java/com/autotune/analyzer/serviceObjects/Converters.java
Update experiment validation logic to work with multiple datasources and enforce that at least one valid datasource is configured for local monitoring.	Change datasource existence validation to iterate over kruizeObject.getDatasources() and validate each name against DataSourceCollection. Return a specific error when no serviceable datasources are configured rather than failing only on a missing single datasource. Update mandatory field validation for local monitoring to accept either the legacy single datasource or a non-empty datasources list.	src/main/java/com/autotune/analyzer/experiment/ExperimentValidation.java
Extend layer detection to run against configured datasources and aggregate layer info per container with robust logging and error handling.	Refactor ServiceHelpers.detectLayers to obtain datasources from CreateExperimentAPIObject.getDatasources() with null/empty guards and logging. Iterate through Kubernetes and container objects, calling LayerUtils.detectLayers with the namespace, container name, and datasources list. Aggregate detected layers into a per-container map; on success, persist it via containerAPIObject.setLayerMap and log counts; log and swallow exceptions per-container to avoid failing the whole flow.	src/main/java/com/autotune/analyzer/utils/ServiceHelpers.java
Make the recommendation engine multi-datasource-aware, requiring a Prometheus datasource for metrics collection, and add Cryostat-specific behavior for PromQL queries and accelerator detection.	Replace simple kruizeObject.getDataSource() usage with logic that selects a datasource whose provider is Prometheus from kruizeObject.getDatasources() or validates the legacy single datasource as Prometheus-only. Throw an exception if no Prometheus datasource can be resolved for an experiment. In fetchMetricsBasedOnProfileAndDatasource, introduce promQLDataSourceInfo which for Cryostat providers is reassigned to a Prometheus datasource from DataSourceCollection for use with PromQL queries; log when none is found. Ensure both container and namespace metric fetchers use promQLDataSourceInfo rather than the original Cryostat datasource for PromQL-based queries. Gate accelerator and accelerator-partition detection logic so it runs only when the underlying datasource provider is Prometheus, effectively skipping those steps for Cryostat-only scenarios.	src/main/java/com/autotune/analyzer/recommendations/engine/RecommendationEngine.java
Align runtime recommendation processing with multi-datasource semantics while keeping legacy behavior when any datasource is present.	Adjust RuntimeRecommendationProcessor.handleRuntimeRecommendations to treat missing datasource as "no data" only when both the legacy single datasource and the datasources list are absent or empty, otherwise proceed. Maintain existing use of container layer maps, which are now populated via the multi-datasource-aware layer detection.	src/main/java/com/autotune/analyzer/recommendations/engine/RuntimeRecommendationProcessor.java
Extend analyzer constants to support pod label usage in presence detection logic.	Add LABEL_POD constant to AnalyzerConstants.RuntimeLayerConstants for future/use in PromQL label-based layer presence detection.	src/main/java/com/autotune/analyzer/utils/AnalyzerConstants.java

---

Tips and commands

Interacting with Sourcery

• Trigger a new review: Comment @sourcery-ai review on the pull request.
• Continue discussions: Reply directly to Sourcery's review comments.
• Generate a GitHub issue from a review comment: Ask Sourcery to create an
  issue from a review comment by replying to it. You can also reply to a
  review comment with @sourcery-ai issue to create an issue from it.
• Generate a pull request title: Write @sourcery-ai anywhere in the pull
  request title to generate a title at any time. You can also comment
  @sourcery-ai title on the pull request to (re-)generate the title at any time.
• Generate a pull request summary: Write @sourcery-ai summary anywhere in
  the pull request body to generate a PR summary at any time exactly where you
  want it. You can also comment @sourcery-ai summary on the pull request to
  (re-)generate the summary at any time.
• Generate reviewer's guide: Comment @sourcery-ai guide on the pull
  request to (re-)generate the reviewer's guide at any time.
• Resolve all Sourcery comments: Comment @sourcery-ai resolve on the
  pull request to resolve all Sourcery comments. Useful if you've already
  addressed all the comments and don't want to see them anymore.
• Dismiss all Sourcery reviews: Comment @sourcery-ai dismiss on the pull
  request to dismiss all existing Sourcery reviews. Especially useful if you
  want to start fresh with a new review - don't forget to comment
  @sourcery-ai review to trigger a new review!

Customizing Your Experience

Access your dashboard to:

• Enable or disable review features such as the Sourcery-generated pull request
  summary, the reviewer's guide, and others.
• Change the review language.
• Add, remove or edit custom review instructions.
• Adjust other review settings.

Getting Help

• Contact our support team for questions or feedback.
• Visit our documentation for detailed guides and information.
• Keep in touch with the Sourcery team by following us on X/Twitter, LinkedIn or GitHub.

[sourcery-ai]

Hey - I've found 1 issue, and left some high level feedback:

• When falling back from a Cryostat datasource to Prometheus in fetchMetricsBasedOnProfileAndDatasource, you currently select the first Prometheus entry from the global DataSourceCollection; consider restricting this to the experiment's configured datasources to avoid silently using an unintended Prometheus instance.
• RuntimeRecommendationProcessor still treats the datasource as effectively singular (only gating on presence of getDataSource/getDatasources); if multiple datasources are configured, it may be worth explicitly selecting or validating which one drives runtime recommendations to avoid ambiguous behavior.
• The datasource validation logic mixes direct field checks (getDataSource() + null checks on getDatasources()) with getDatasources()'s backward-compatible behavior; consider standardizing on getDatasources() everywhere to simplify the conditions and avoid subtle differences between validation paths.

Prompt for AI Agents

Please address the comments from this code review:

## Overall Comments
- When falling back from a Cryostat datasource to Prometheus in `fetchMetricsBasedOnProfileAndDatasource`, you currently select the first Prometheus entry from the global `DataSourceCollection`; consider restricting this to the experiment's configured datasources to avoid silently using an unintended Prometheus instance.
- RuntimeRecommendationProcessor still treats the datasource as effectively singular (only gating on presence of `getDataSource`/`getDatasources`); if multiple datasources are configured, it may be worth explicitly selecting or validating which one drives runtime recommendations to avoid ambiguous behavior.
- The datasource validation logic mixes direct field checks (`getDataSource()` + null checks on `getDatasources()`) with `getDatasources()`'s backward-compatible behavior; consider standardizing on `getDatasources()` everywhere to simplify the conditions and avoid subtle differences between validation paths.

## Individual Comments

### Comment 1
<location path="src/main/java/com/autotune/analyzer/serviceObjects/CreateExperimentAPIObject.java" line_range="177-178" />
<code_context>
+     */
+    public List<String> getDatasources() {
+        // Backward compatibility: if datasources is null but datasource is set
+        if (datasources == null && datasource != null) {
+            return Arrays.asList(datasource);
+        }
</code_context>
<issue_to_address>
**issue (bug_risk):** Using `List.of` for backward compatibility may cause issues on Java 8 runtimes.

`KruizeObject#getDatasources` already uses `Arrays.asList(datasource)` for this backward-compatibility case. For consistency and to avoid Java 8 compilation issues, use `Collections.singletonList(datasource)` or `Arrays.asList(datasource)` instead of `List.of(datasource)` here.
</issue_to_address>

---

Sourcery is free for open source - if you like our reviews please consider sharing them ✨

• X
• Mastodon
• LinkedIn
• Facebook

_{Help me be more useful! Please click 👍 or 👎 on each comment and I'll use the feedback to improve your reviews.}