perf(finding): drop four dojo_finding indexes that no query uses

dojo/db_migrations/0295_drop_unused_finding_indexes.py

@@ -0,0 +1,79 @@
+"""Drop four `dojo_finding` indexes that no query uses.
+
+`dojo_finding` is the busiest table in the schema and carries close to fifty indexes, every
+one of which is maintained on each row written. Import is where that is felt: a scan
+creating thousands of findings pays for all of them per row.
+
+These four were measured as completely unused -- `pg_stat_user_indexes.idx_scan = 0` -- on
+four long-running production instances, three of which had accumulated statistics for about
+seventeen months:
+
+    (epss_percentile)      the largest of them; note (epss_score) IS used, so findings are
+                           sorted by score and the percentile is stored and displayed but is
+                           never an access path
+    (line)
+    (known_exploited)
+    idx_finding_sev_open_unver   partial: (severity, -numerical_severity)
+                                 WHERE active AND NOT verified
+
+Indexes that were unused on *some* instances but not others were deliberately left alone --
+they track optional features, and dropping on that evidence would break the installs that do
+use them.
+
+Built with DROP INDEX CONCURRENTLY (non-atomic migration, following 0280) so it takes no
+exclusive lock on a large `dojo_finding`. IF EXISTS makes it idempotent, and the reverse
+rebuilds concurrently, so a downgrade does not lock the table either.
+"""
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+    atomic = False
+
+    dependencies = [
+        ("dojo", "0294_usercontactinfo_language"),
+    ]
+
+    operations = [
+        # RunSQL does the concurrent DB work; state_operations keeps Django's model state in
+        # step with the Meta change, so a later makemigrations stays a no-op.
+        migrations.SeparateDatabaseAndState(
+            database_operations=[
+                migrations.RunSQL(
+                    sql="DROP INDEX CONCURRENTLY IF EXISTS dojo_findin_epss_pe_567499_idx",
+                    reverse_sql=(
+                        "CREATE INDEX CONCURRENTLY IF NOT EXISTS dojo_findin_epss_pe_567499_idx "
+                        "ON dojo_finding (epss_percentile)"
+                    ),
+                ),
+                migrations.RunSQL(
+                    sql="DROP INDEX CONCURRENTLY IF EXISTS dojo_findin_line_fea329_idx",
+                    reverse_sql=(
+                        "CREATE INDEX CONCURRENTLY IF NOT EXISTS dojo_findin_line_fea329_idx "
+                        "ON dojo_finding (line)"
+                    ),
+                ),
+                migrations.RunSQL(
+                    sql="DROP INDEX CONCURRENTLY IF EXISTS dojo_findin_known_e_8c584e_idx",
+                    reverse_sql=(
+                        "CREATE INDEX CONCURRENTLY IF NOT EXISTS dojo_findin_known_e_8c584e_idx "
+                        "ON dojo_finding (known_exploited)"
+                    ),
+                ),
+                migrations.RunSQL(
+                    sql="DROP INDEX CONCURRENTLY IF EXISTS idx_finding_sev_open_unver",
+                    reverse_sql=(
+                        "CREATE INDEX CONCURRENTLY IF NOT EXISTS idx_finding_sev_open_unver "
+                        "ON dojo_finding (severity, numerical_severity DESC) "
+                        "WHERE (active AND NOT verified)"
+                    ),
+                ),
+            ],
+            state_operations=[
+                migrations.RemoveIndex(model_name="finding", name="dojo_findin_epss_pe_567499_idx"),
+                migrations.RemoveIndex(model_name="finding", name="dojo_findin_line_fea329_idx"),
+                migrations.RemoveIndex(model_name="finding", name="dojo_findin_known_e_8c584e_idx"),
+                migrations.RemoveIndex(model_name="finding", name="idx_finding_sev_open_unver"),
+            ],
+        ),
+    ]

dojo/finding/models.py

@@ -469,7 +469,6 @@ class Meta:
 
             models.Index(fields=["cve"]),
             models.Index(fields=["epss_score"]),
-            models.Index(fields=["epss_percentile"]),
             models.Index(fields=["cwe"]),
             models.Index(fields=["out_of_scope"]),
             models.Index(fields=["false_p"]),
@@ -482,12 +481,10 @@ class Meta:
             models.Index(fields=["hash_code"]),
             models.Index(fields=["unique_id_from_tool"]),
             # models.Index(fields=['file_path']), # can't add index because the field has max length 4000.
-            models.Index(fields=["line"]),
             models.Index(fields=["component_name"]),
             models.Index(fields=["duplicate"]),
             models.Index(fields=["is_mitigated"]),
             models.Index(fields=["duplicate_finding", "id"]),
-            models.Index(fields=["known_exploited"]),
             models.Index(fields=["ransomware_used"]),
             models.Index(fields=["kev_date"]),
             models.Index(
@@ -514,11 +511,6 @@ class Meta:
                 name="idx_finding_open_active_sev",
                 condition=models.Q(active=True, is_mitigated=False),
             ),
-            models.Index(
-                fields=["severity", "-numerical_severity"],
-                name="idx_finding_sev_open_unver",
-                condition=models.Q(active=True, verified=False),
-            ),
             models.Index(
                 fields=["test", "sla_expiration_date", "date"],
                 name="idx_finding_sla_breach_cov",