fix: Use original group_results api with optimized indexes (#8953)

Co-authored-by: Cameron Dawson <[email protected]>

Author: camd

treeherder/log_parser/failureline.py

@@ -193,11 +193,7 @@ def replace_astral(log_list):
         yield item
 
 
-def get_group_results_legacy(repository, push):
-    """
-    Legacy implementation - preserved for testing and comparison.
-    Performance: ~3.3 seconds average.
-    """
+def get_group_results(repository, push):
     groups = Group.objects.filter(
         job_logs__job__push__revision=push.revision,
         job_logs__job__push__repository=repository,
@@ -215,63 +211,3 @@ def get_group_results_legacy(repository, push):
         )
 
     return by_task_id
-
-
-def get_group_results(repository, push):
-    """
-    OPTIMIZED IMPLEMENTATION - Best performing non-cached query.
-
-    Performance: ~0.129s (27% faster than legacy 0.176s)
-
-    This implementation uses an optimized SQL query starting from the job table
-    for the best join performance with the new database indexes.
-
-    RECOMMENDED DATABASE INDEXES (already applied):
-
-    -- Composite indexes for optimal query performance
-    CREATE INDEX CONCURRENTLY idx_group_status_composite
-        ON group_status(status, job_log_id, group_id);
-
-    CREATE INDEX CONCURRENTLY idx_job_push_id
-        ON job(push_id);
-
-    CREATE INDEX CONCURRENTLY idx_job_log_job_id
-        ON job_log(job_id);
-
-    CREATE INDEX CONCURRENTLY idx_taskcluster_metadata_job_id
-        ON taskcluster_metadata(job_id);
-
-    CREATE INDEX CONCURRENTLY idx_push_revision_repo
-        ON push(revision, repository_id);
-    """
-    from django.db import connection
-
-    ok_status = GroupStatus.OK
-
-    query = """
-        SELECT
-            tcm.task_id,
-            g.name,
-            gs.status
-        FROM job j
-        INNER JOIN taskcluster_metadata tcm ON j.id = tcm.job_id
-        INNER JOIN job_log jl ON j.id = jl.job_id
-        INNER JOIN group_status gs ON jl.id = gs.job_log_id
-        INNER JOIN "group" g ON gs.group_id = g.id
-        WHERE j.push_id = %s
-        AND gs.status IN (%s, %s)
-        ORDER BY tcm.task_id
-    """
-
-    by_task_id = {}
-
-    with connection.cursor() as cursor:
-        cursor.execute(query, [push.id, GroupStatus.OK, GroupStatus.ERROR])
-        rows = cursor.fetchall()
-
-        for task_id, group_name, status in rows:
-            if task_id not in by_task_id:
-                by_task_id[task_id] = {}
-            by_task_id[task_id][group_name] = status == ok_status
-
-    return by_task_id

treeherder/webapp/api/push.py

@@ -10,10 +10,7 @@
 from rest_framework.response import Response
 from rest_framework.status import HTTP_400_BAD_REQUEST, HTTP_404_NOT_FOUND
 
-from treeherder.log_parser.failureline import (
-    get_group_results,
-    get_group_results_legacy,
-)
+from treeherder.log_parser.failureline import get_group_results
 from treeherder.model.models import Commit, Job, JobType, Push, Repository
 from treeherder.push_health.builds import get_build_failures
 from treeherder.push_health.compare import get_commit_history
@@ -493,9 +490,6 @@ def report_if_short_revision(self, param, revision):
     def group_results(self, request, project):
         """
         Return the results of all the test groups for this push.
-
-        OPTIMIZED IMPLEMENTATION - Best performing non-cached query.
-        Performance: ~0.129s (27% faster than legacy 0.176s)
         """
         revision = request.query_params.get("revision")
 
@@ -507,20 +501,3 @@ def group_results(self, request, project):
 
         groups = get_group_results(repository, push)
         return Response(groups)
-
-    @action(detail=False)
-    def group_results_legacy(self, request, project):
-        """
-        LEGACY ENDPOINT - Original implementation preserved for comparison.
-        Performance: ~0.176s (baseline for comparison)
-        """
-        revision = request.query_params.get("revision")
-
-        try:
-            repository = Repository.objects.get(name=project)
-            push = Push.objects.get(revision=revision, repository=repository)
-        except Push.DoesNotExist:
-            return Response(f"No push with revision: {revision}", status=HTTP_404_NOT_FOUND)
-
-        groups = get_group_results_legacy(repository, push)
-        return Response(groups)