fix(services): eliminate n+1 query in unit detail endpoint

[mhieta]

Problem

Production users were experiencing performance degradation on the /v2/unit/{pk}/ API endpoint due to an N+1 query issue. The endpoint was executing up to 64 repeated database queries to the services_service table when retrieving a single unit, causing slow response times and increased database load.

Affected endpoint: /v2/unit/{pk}/

Root Cause

The UnitViewSet.retrieve() method bypassed the optimized get_queryset() method, which contains essential prefetch_related calls for related objects. Instead, it directly called _get_unit(pk), which performed a simple Unit.objects.get() query without any prefetching.

As a result, when the UnitSerializer accessed related fields during serialization (such as service_nodes, connections, accessibility_properties, and keywords), each access triggered a separate database query, creating the classic N+1 query problem.

Technical details:

• The retrieve method bypassed get_queryset's prefetching logic
• _get_unit() performed direct database queries: Unit.objects.get(pk=pk, public=True, is_active=True)
• During serialization, UnitSerializer.to_representation() accessed related fields using .all() calls
• Each related object access resulted in a new query to the database

Solution

Refactored the UnitViewSet.retrieve() method to utilize the optimized queryset from get_queryset(), ensuring that all prefetch_related optimizations are applied before fetching the unit.

Implementation:

1. Modified UnitViewSet.retrieve() to call get_queryset() and pass the optimized queryset to _get_unit()
2. Updated _get_unit() signature to accept an optional queryset parameter
3. Changed query execution in _get_unit() from Unit.objects.get() to queryset.get()
4. Maintained backward compatibility with fallback logic for UnitAlias lookups

The solution preserves all existing functionality while leveraging Django ORM's prefetch capabilities that were already configured in get_queryset().

Performance Impact

Before

• Database Queries: 64+ repeated queries to services_service table
• Affected Endpoint: /v2/unit/{pk}/

After

• Database Queries: < 30 total queries (>53% reduction)
• Response Time: Significantly improved
• Database Load: Reduced by eliminating redundant queries
• Scalability: Better performance under high load

Measured Improvements

• Query count for unit detail requests: 64 → <30 (>50% reduction)
• All related objects prefetched in single pass
• No additional queries during serialization

References

• Jira Ticket: Refs: PL-218
• Issue Type: Performance - N+1 Database Queries

Impact Assessment

This fix will:

1. Improve API response times for unit detail requests
2. Reduce database server load by eliminating redundant queries
3. Enhance user experience with faster page loads
4. Improve system scalability under high traffic

The change is backward compatible and does not affect API behavior or responses - only internal query optimization.

Additional Notes

• The get_queryset() method already contained all necessary prefetch_related calls
• This fix simply ensures those optimizations are actually used during single-unit retrieval
• The solution follows Django best practices for ORM optimization
• No changes required to serializers or models
• Fix is minimal and focused on the root cause

[azure-pipelines]

PALVELUKARTTA-API branch is deployed to platta: https://palvelukartta-api-pr378.dev.hel.ninja 🚀🚀🚀

[azure-pipelines]

PALVELUKARTTA-API branch is deployed to platta: https://palvelukartta-api-pr378.dev.hel.ninja 🚀🚀🚀

[mhieta]

@sentry review

[Copilot]

Pull request overview

This pull request fixes an N+1 query performance issue in the unit detail API endpoint (/v2/unit/{pk}/) that was causing up to 64+ database queries per request. The fix ensures that the retrieve() method uses the optimized queryset with prefetch operations, reducing queries by over 50% while maintaining backward compatibility with UnitAlias lookups.

Changes:

• Modified UnitViewSet.retrieve() to call get_queryset() and pass the optimized queryset to _get_unit()
• Updated _get_unit() to accept an optional queryset parameter and use it for database operations
• Added comprehensive test coverage to prevent regression of the N+1 query issue

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 3 comments.

File	Description
services/api.py	Refactored _get_unit() and retrieve() methods to use prefetched queryset, eliminating N+1 queries while preserving UnitAlias functionality
services/tests/test_unit_view_set_api.py	Added four new test cases covering N+1 prevention, include parameters, alias access, and filter enforcement

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated no new comments.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[azure-pipelines]

PALVELUKARTTA-API branch is deployed to platta: https://palvelukartta-api-pr378.dev.hel.ninja 🚀🚀🚀

[sonarqubecloud]

Quality Gate passed

Issues
0 New issues
0 Accepted issues

Measures
0 Security Hotspots
100.0% Coverage on New Code
0.0% Duplication on New Code

See analysis details on SonarQube Cloud

[azure-pipelines]

PALVELUKARTTA-API branch is deployed to platta: https://palvelukartta-api-pr378.dev.hel.ninja 🚀🚀🚀