fix: docker cache issues, dashboards

[MasterPtato]

Description

Please include a summary of the changes and the related issue. Please also include relevant motivation and context.

Type of change

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds functionality)
• Breaking change (fix or feature that would cause existing functionality to not work as expected)
• This change requires a documentation update

How Has This Been Tested?

Please describe the tests that you ran to verify your changes.

Checklist:

• My code follows the style guidelines of this project
• I have performed a self-review of my code
• I have commented my code, particularly in hard-to-understand areas
• I have made corresponding changes to the documentation
• My changes generate no new warnings
• I have added tests that prove my fix is effective or that my feature works
• New and existing unit tests pass locally with my changes

[railway-app]

🚅 Deployed to the rivet-pr-4181 environment in rivet-frontend

Service	Status	Web	Updated (UTC)
website	😴 Sleeping (View Logs)	Web	Feb 13, 2026 at 7:57 pm
frontend-inspector	❌ Build Failed (View Logs)	Web	Feb 11, 2026 at 9:22 pm
frontend-cloud	❌ Build Failed (View Logs)	Web	Feb 11, 2026 at 9:21 pm
mcp-hub	✅ Success (View Logs)	Web	Feb 11, 2026 at 9:20 pm
ladle	❌ Build Failed (View Logs)	Web	Feb 11, 2026 at 9:19 pm

[MasterPtato]

• fix: add max prunes to pruner #4191
• fix: add max param to prune signals cmd #4190
• fix: reduce metrics aggregator cardinality #4183
• fix: docker cache issues, dashboards #4181 👈 (View in Graphite)
• main

---

How to use the Graphite Merge Queue

Add the label merge-queue to this PR to add it to the merge queue.

You must have a Graphite account in order to use the merge queue. Sign up using this link.

An organization admin has enabled the Graphite Merge Queue in this repository.

Please do not merge from GitHub as this will restart CI on PRs being processed by the merge queue.

This stack of pull requests is managed by Graphite. Learn more about stacking.

[claude]

Code Review

This PR contains several fixes for Docker cache issues and dashboard improvements. Here's my analysis:

✅ Positive Changes

1. Critical Bug Fix in Gasoline Worker Distribution (gasoline/src/db/kv/mod.rs)

-let active_worker_count = active_worker_ids.len() as u64;
+let active_worker_count = active_worker_ids.len().max(1) as u64;

Excellent fix. This prevents division by zero when active_worker_ids is empty. The .max(1) ensures there's always at least one worker for workload distribution calculations.

2. Docker Build Cache Fix (util/build.rs)

+println!("cargo:rerun-if-env-changed=OVERRIDE_GIT_SHA");

Good addition. This directive tells Cargo to rerun the build script if the OVERRIDE_GIT_SHA environment variable changes, which is crucial for Docker caching behavior. This ensures the build cache is invalidated when the git SHA override changes between builds.

3. Debugging Output (docker/universal/Dockerfile)

+echo "Copying binary" && \

Minor but helpful for debugging Docker build steps during development.

4. Grafana Dashboard Improvements (grafana-dashboards/*.json)

The dashboard changes include:

• Reorganized panels with better spacing and positioning
• New heatmap visualizations for CPU usage and load shedding
• Improved metric queries with better grouping logic
• Fixed panel positions (y-coordinates updated to accommodate new panels)
• Added new workflow dispatch rate metrics
• Better visualization options (stacking, fill opacity)

---

🔍 Areas for Consideration

1. Pegboard Dashboard Query Complexity

The Pegboard dashboard now uses more complex Prometheus queries with nested joins:

sum by (rivet_datacenter) (
    max by(rivet_project, rivet_datacenter) (
        rivet_pegboard_actor_pending_allocation{...}
        * on(k8s_pod_name) group_left(rivet_project, rivet_datacenter)
        (
            (rivet_gasoline_worker_last_metrics_publish{...} == bool
            on(rivet_project, rivet_datacenter) group_left()
            max by (rivet_project, rivet_datacenter) (rivet_gasoline_worker_last_metrics_publish{...}))
        )
    )
)

Question: What problem does this solve? The nested query appears to filter metrics to only show data from workers that most recently published metrics. While this makes sense for accuracy, consider:

• Performance impact on Prometheus with high cardinality
• Query timeout risk in large deployments
• Adding a comment explaining why this complex join is necessary

2. Missing PR Description

The PR template checkboxes are empty. Could you fill these in?

• What specific Docker cache issues were fixed?
• What testing was performed?
• Are there any breaking changes?

3. Dashboard Version Bumps

• Gasoline dashboard: v8 → v9
• Pegboard dashboard: v1 → v2

These look like auto-generated version increments from Grafana, but it would be good to document what specific visualizations were added/changed in the commit message or PR description.

---

📋 Recommendations

1. Update PR Description: Fill in the template with:

   • Summary of the Docker cache fix
   • Explanation of the worker count division-by-zero fix
   • Description of dashboard improvements
   • Testing performed

2. Consider Performance Testing: The new Pegboard Prometheus queries are significantly more complex. If possible, test these on a staging environment with realistic data volumes.

3. Optional - Code Comment: Consider adding a comment in mod.rs:1177 explaining why .max(1) is needed:

   // Ensure at least 1 worker to prevent division by zero in workload distribution
   let active_worker_count = active_worker_ids.len().max(1) as u64;

4. Dashboard Documentation: Consider adding comments in the Grafana JSON (if Grafana preserves them) or in a separate doc explaining the complex query logic for future maintenance.

---

✅ Summary

Overall, this is a solid bug fix PR that addresses:

• A potential division-by-zero issue
• Docker cache invalidation problems
• Improved observability through better dashboards

The code changes are minimal, focused, and correct. The main areas for improvement are documentation and testing validation rather than code quality issues.

Recommendation: Approve after PR description is filled out.

[graphite-app]

Legend format references {{signal_name}} but the query groups by sub_workflow_name. This mismatch will result in empty or incorrect legend labels.

Fix:

"legendFormat": "{{sub_workflow_name}}",

Suggested change

	"legendFormat": "{{signal_name}}",
	"legendFormat": "{{sub_workflow_name}}",

Spotted by Graphite Agent



Is this helpful? React 👍 or 👎 to let us know.

[graphite-app]

Format is set to "heatmap" but the panel type is "timeseries" (line 1724). This format mismatch will cause visualization errors.

Fix:

"format": "time_series",

Or remove the format field entirely to use the default for timeseries panels.

Suggested change

	"format": "heatmap",
	"format": "time_series",

Spotted by Graphite Agent



Is this helpful? React 👍 or 👎 to let us know.