Update migrations list command to show migrations that no longer exist in the codebase (#6299)

Fixes #2159

Running `snuba migrations list` will now show migrations that are in
clickhouse but no longer in the codebase (e.g. because snuba was
downgraded).

# Testing
Deleted migration in codebase to check that it shows up in the list.
<img width="668" alt="image"
src="https://github.com/user-attachments/assets/90f855b9-94af-416d-9452-0335c2abd8d6">

Author: davidtsuk

snuba/admin/clickhouse/migration_checks.py

@@ -100,7 +100,7 @@ def __init__(
         ).get_migrations()
 
         migration_statuses = {}
-        for migration_id, status, _ in migrations:
+        for migration_id, status, _, _ in migrations:
             migration_statuses[migration_id] = {
                 "migration_id": migration_id,
                 "status": status,

snuba/admin/views.py

@@ -208,7 +208,7 @@ def migrations_groups_list(group: str) -> Response:
                             "status": status.value,
                             "blocking": blocking,
                         }
-                        for migration_id, status, blocking in runner_group_migrations
+                        for migration_id, status, blocking, _ in runner_group_migrations
                     ]
                 ),
                 200,

snuba/cli/migrations.py

@@ -37,10 +37,10 @@ def list() -> None:
     setup_logging()
     check_clickhouse_connections(CLUSTERS)
     runner = Runner()
-    for group, group_migrations in runner.show_all():
+    for group, group_migrations in runner.show_all(include_nonexistent=True):
         readiness_state = get_group_readiness_state(group)
         click.echo(f"{group.value} (readiness_state: {readiness_state.value})")
-        for migration_id, status, blocking in group_migrations:
+        for migration_id, status, blocking, existing in group_migrations:
             symbol = {
                 Status.COMPLETED: "X",
                 Status.NOT_STARTED: " ",
@@ -53,7 +53,11 @@ def list() -> None:
             if status != Status.COMPLETED and blocking:
                 blocking_text = " (blocking)"
 
-            click.echo(f"[{symbol}]  {migration_id}{in_progress_text}{blocking_text}")
+            existing_text = "" if existing else " (this migration no longer exists)"
+
+            click.echo(
+                f"[{symbol}]  {migration_id}{in_progress_text}{blocking_text}{existing_text}"
+            )
 
         click.echo()
 

snuba/migrations/runner.py

@@ -1,3 +1,4 @@
+from collections import defaultdict
 from datetime import datetime
 from functools import partial
 from typing import List, Mapping, MutableMapping, NamedTuple, Optional, Sequence, Tuple
@@ -64,6 +65,7 @@ class MigrationDetails(NamedTuple):
     migration_id: str
     status: Status
     blocking: bool
+    exists: bool
 
 
 class Runner:
@@ -133,7 +135,7 @@ def force_overwrite_status(
         )
 
     def show_all(
-        self, groups: Optional[Sequence[str]] = None
+        self, groups: Optional[Sequence[str]] = None, include_nonexistent: bool = False
     ) -> List[Tuple[MigrationGroup, List[MigrationDetails]]]:
         """
         Returns the list of migrations and their statuses for each group.
@@ -148,6 +150,9 @@ def show_all(
             migration_groups = get_active_migration_groups()
 
         migration_status = self._get_migration_status(migration_groups)
+        clickhouse_group_migrations = defaultdict(set)
+        for group, migration_id in migration_status.keys():
+            clickhouse_group_migrations[group].add(migration_id)
 
         def get_status(migration_key: MigrationKey) -> Status:
             return migration_status.get(migration_key, Status.NOT_STARTED)
@@ -156,15 +161,31 @@ def get_status(migration_key: MigrationKey) -> Status:
             group_migrations: List[MigrationDetails] = []
             group_loader = get_group_loader(group)
 
-            for migration_id in group_loader.get_migrations():
+            migration_ids = group_loader.get_migrations()
+            for migration_id in migration_ids:
                 migration_key = MigrationKey(group, migration_id)
                 migration = group_loader.load_migration(migration_id)
                 group_migrations.append(
                     MigrationDetails(
-                        migration_id, get_status(migration_key), migration.blocking
+                        migration_id,
+                        get_status(migration_key),
+                        migration.blocking,
+                        True,
                     )
                 )
 
+            if include_nonexistent:
+                non_existing_migrations = clickhouse_group_migrations.get(
+                    group, set()
+                ).difference(set(migration_ids))
+                for migration_id in non_existing_migrations:
+                    migration_key = MigrationKey(group, migration_id)
+                    group_migrations.append(
+                        MigrationDetails(
+                            migration_id, get_status(migration_key), False, False
+                        )
+                    )
+
             migrations.append((group, group_migrations))
 
         return migrations

tests/admin/clickhouse_migrations/test_migration_checks.py

@@ -29,9 +29,9 @@ def group_loader() -> GroupLoader:
 
 
 RUN_MIGRATIONS: Sequence[MigrationDetails] = [
-    MigrationDetails("0001", Status.COMPLETED, True),
-    MigrationDetails("0002", Status.NOT_STARTED, True),
-    MigrationDetails("0003", Status.NOT_STARTED, True),
+    MigrationDetails("0001", Status.COMPLETED, True, True),
+    MigrationDetails("0002", Status.NOT_STARTED, True, True),
+    MigrationDetails("0003", Status.NOT_STARTED, True, True),
 ]
 
 
@@ -62,9 +62,9 @@ def test_status_checker_run(
 
 
 REVERSE_MIGRATIONS: Sequence[MigrationDetails] = [
-    MigrationDetails("0001", Status.COMPLETED, True),
-    MigrationDetails("0002", Status.IN_PROGRESS, True),
-    MigrationDetails("0003", Status.NOT_STARTED, True),
+    MigrationDetails("0001", Status.COMPLETED, True, True),
+    MigrationDetails("0002", Status.IN_PROGRESS, True, True),
+    MigrationDetails("0003", Status.NOT_STARTED, True, True),
 ]
 
 
@@ -155,7 +155,10 @@ def test_run_migration_checks_and_policies(
     mock_policy = Mock()
     checker = mock_checker()
     mock_runner.show_all.return_value = [
-        (MigrationGroup("events"), [MigrationDetails("0001", Status.COMPLETED, True)])
+        (
+            MigrationGroup("events"),
+            [MigrationDetails("0001", Status.COMPLETED, True, True)],
+        )
     ]
 
     mock_policy.can_run.return_value = policy_result[0]

tests/migrations/test_runner.py

@@ -110,6 +110,26 @@ def test_show_all_for_groups() -> None:
     assert all([migration.status == Status.COMPLETED for migration in migrations])
 
 
+@pytest.mark.clickhouse_db
+def test_show_all_nonexistent_migration() -> None:
+    runner = Runner()
+    assert all(
+        [
+            migration.status == Status.NOT_STARTED
+            for (_, group_migrations) in runner.show_all()
+            for migration in group_migrations
+        ]
+    )
+    runner.run_all(force=True)
+    assert all(
+        [
+            migration.status == Status.COMPLETED
+            for (_, group_migrations) in runner.show_all()
+            for migration in group_migrations
+        ]
+    )
+
+
 @pytest.mark.clickhouse_db
 def test_run_migration() -> None:
     runner = Runner()