[OPIK-7256] fix(backend): make project size-estimate cache actually engage

[thiagohora]

Details

The per-project size estimate (WorkspaceMetadataDAO.getProjectMetadata — a 1000-span size sample plus uniq(id) over the whole project) is meant to be cached, but the cache never engaged, so the query ran on essentially every traces/spans find/search request and became one of the top read-volume queries on the ClickHouse cluster (~65 executions/min). Two stacked defects:

• @Cacheable(name = "project_metadata", ...) sits on a private method, and Guice method interception (which implements @Cacheable) cannot intercept private methods — the annotation was silently dead. Made the method package-private with a comment explaining why it must not be private.
• The cache name project_metadata was absent from cacheManager.caches; the config only carried an orphaned workspace_metadata entry that no code references (stale name), so even an intercepted call would have used the 1-second defaultDuration. Renamed the entry to project_metadata (env CACHE_MANAGER_PROJECT_METADATA_DURATION, default PT1H, matching the working experiment_metadata sibling).

With the cache engaged, per-(workspace, project) executions drop from once per request to once per TTL window.

Change checklist

• User facing
• Documentation update

Issues

• Resolves #
• OPIK-7256

AI-WATERMARK

AI-WATERMARK: yes

• If yes:
  • Tools: Claude Code
  • Model(s): Claude Fable 5
  • Scope: visibility fix, config rename, commit and PR text
  • Human verification: pending reviewer; production query-frequency numbers measured read-only from system.query_log

Testing

• Added WorkspaceMetadataServiceCacheTest: boots the app harness with a counting-stub WorkspaceMetadataDAO and asserts a repeated getProjectMetadata call for the same key is served from the cache (DAO hit exactly once), and a different project misses. The test fails if the method becomes non-interceptable again (e.g. private) or if the project_metadata cache name is dropped from configuration.
• mvn compile and mvn test-compile pass; mvn spotless:apply produced no further changes
• Guice interception limitation is documented behavior (private methods are never intercepted); the sibling experiment_metadata cache — public method + configured name — demonstrates the working pattern this change aligns with
• Production system.query_log (7 days): ~657K executions/week of this query with per-request cardinality, confirming the cache was not engaging

🤖 Generated with Claude Code

Documentation

No user-facing documentation impact. The new env var CACHE_MANAGER_PROJECT_METADATA_DURATION (replacing the orphaned CACHE_MANAGER_WORKSPACE_METADATA_DURATION) is documented inline in config.yml.

Note on the sample query itself: we evaluated also reworking the sample CTE's ORDER BY (sorting key) DESC LIMIT 1000 (it reads the whole project because reverse-order read never engages for this shape), but dropping the ORDER BY would pin the sample to the oldest ~1000 spans instead of the latest, so the estimate would stop tracking current span sizes. Keeping the query as-is; with the cache engaged it runs at most once per TTL per project.

[github-actions]

⏱️ pre-commit per-hook timing

Hook	Description	Result	Duration
☕ spotless — java backend	Format Java code	✅	5.39s
Total (1 ran)			5.39s

⏭️ 39 skipped (no matching files changed)

Hook	Description	Result
🐍 trim trailing whitespace — python sdk	Strip trailing whitespace	⏭️
🐍 fix end of files — python sdk	Ensure files end in a newline	⏭️
🐍 ruff — python sdk	Lint + autofix Python (ruff)	⏭️
🐍 ruff-format — python sdk	Format Python code (ruff)	⏭️
🐍 mypy — python sdk	Static type check	⏭️
🤖 trim trailing whitespace — optimizer	Strip trailing whitespace	⏭️
🤖 fix end of files — optimizer	Ensure files end in a newline	⏭️
🤖 check yaml — optimizer	Validate YAML syntax	⏭️
🤖 check json — optimizer	Validate JSON syntax	⏭️
🤖 check toml — optimizer	Validate TOML syntax	⏭️
🤖 check for added large files — optimizer	Block large files (>1MB)	⏭️
🔐 detect private key — optimizer	Block committed private keys	⏭️
🤖 check for merge conflicts — optimizer	Block merge-conflict markers	⏭️
🤖 check for case conflicts — optimizer	Block case-only name clashes	⏭️
🤖 pyupgrade — optimizer	Modernize Python syntax	⏭️
🤖 ruff — optimizer	Lint + autofix Python (ruff)	⏭️
🤖 ruff-format — optimizer	Format Python code (ruff)	⏭️
🤖 mypy — optimizer	Static type check	⏭️
📓 nbstripout — optimizer notebooks	Strip notebook output	⏭️
📝 markdownlint — optimizer	Lint Markdown	⏭️
🔤 codespell — optimizer	Fix common misspellings	⏭️
📊 radon cc — optimizer	Cyclomatic-complexity gate	⏭️
📊 radon raw — optimizer	Raw size metrics gate	⏭️
📊 xenon — optimizer	Fail on complexity thresholds	⏭️
📊 lizard — optimizer	Cyclomatic-complexity gate	⏭️
🧹 vulture — optimizer	Find dead code	⏭️
🛡️ trim trailing whitespace — guardrails	Strip trailing whitespace	⏭️
🛡️ fix end of files — guardrails	Ensure files end in a newline	⏭️
🛡️ ruff — guardrails	Lint + autofix Python (ruff)	⏭️
🛡️ ruff-format — guardrails	Format Python code (ruff)	⏭️
🛡️ mypy — guardrails	Static type check	⏭️
⚓ helm-docs	Regenerate Helm chart README	⏭️
block non-public FE plugins	Block non-public FE plugins	⏭️
🧪 pre-commit wrapper smoke tests	Self-test the wrapper scripts	⏭️
🌐 eslint — frontend	Lint + autofix JS/TS	⏭️
🌐 typecheck — frontend	Whole-project tsc type check	⏭️
📘 eslint — typescript sdk	Lint + autofix JS/TS	⏭️
📘 typecheck — typescript sdk	Whole-project tsc type check	⏭️
⚙️ actionlint — github workflows	Lint GitHub Actions workflows	⏭️