fix(migrations): Make Snuba event migration more robust (#16364)

Fixes getsentry/self-hosted#322

src/sentry/migrations/0024_auto_20191230_2052.py

@@ -1,6 +1,6 @@
 # -*- coding: utf-8 -*-
 # Generated by Django 1.9.13 on 2019-12-30 20:52
-from __future__ import unicode_literals
+from __future__ import unicode_literals, print_function
 
 import os
 from datetime import timedelta, datetime
@@ -20,9 +20,9 @@ def backfill_eventstream(apps, schema_editor):
     from sentry import eventstore, eventstream
     from sentry.utils.query import RangeQuerySetWrapper
 
-    Event = apps.get_model('sentry', 'Event')
-    Group = apps.get_model('sentry', 'Group')
-    Project = apps.get_model('sentry', 'Project')
+    Event = apps.get_model("sentry", "Event")
+    Group = apps.get_model("sentry", "Group")
+    Project = apps.get_model("sentry", "Project")
 
     # Kill switch to skip this migration
     skip_backfill = os.environ.get("SENTRY_SKIP_EVENTS_BACKFILL_FOR_10", False)
@@ -34,20 +34,26 @@ def backfill_eventstream(apps, schema_editor):
     def get_events(last_days):
         to_date = datetime.now()
         from_date = to_date - timedelta(days=last_days)
-        return Event.objects.filter(datetime__gte=from_date, datetime__lte=to_date, group_id__isnull=False)
+        return Event.objects.filter(
+            datetime__gte=from_date, datetime__lte=to_date, group_id__isnull=False
+        )
 
     def _attach_related(_events):
-        project_ids = {event.project_id for event in _events}
+        project_ids = set()
+        group_ids = set()
+        for event in _events:
+            project_ids.add(event.project_id)
+            group_ids.add(event.group_id)
         projects = {p.id: p for p in Project.objects.filter(id__in=project_ids)}
-        group_ids = {event.group_id for event in _events}
         groups = {g.id: g for g in Group.objects.filter(id__in=group_ids)}
+
         for event in _events:
-            event.project = projects[event.project_id]
-            event.group = groups[event.group_id]
+            event.project = projects.get(event.project_id)
+            event.group = groups.get(event.group_id)
         eventstore.bind_nodes(_events, "data")
 
     if skip_backfill:
-        print("Skipping backfill\n")
+        print("Skipping backfill.\n")
         return
 
     events = get_events(retention_days)
@@ -59,8 +65,13 @@ def _attach_related(_events):
 
     print("Events to process: {}\n".format(count))
 
+    processed = 0
     for event in RangeQuerySetWrapper(events, step=100, callbacks=(_attach_related,)):
         primary_hash = event.get_primary_hash()
+        if event.project is None or event.group is None:
+            print("Skipped {} as group or project information is invalid.\n".format(event))
+            continue
+
         eventstream.insert(
             group=event.group,
             event=event,
@@ -70,8 +81,9 @@ def _attach_related(_events):
             primary_hash=primary_hash,
             skip_consume=True,
         )
+        processed += 1
 
-    print("Done.\n")
+    print("Event migration done. Processed {} of {} events.\n".format(processed, count))
 
 
 class Migration(migrations.Migration):
@@ -89,9 +101,8 @@ class Migration(migrations.Migration):
     # - Adding columns to highly active tables, even ones that are NULL.
     is_dangerous = True
 
-
     dependencies = [
-        ('sentry', '0023_hide_environment_none_20191126'),
+        ("sentry", "0023_hide_environment_none_20191126"),
     ]
 
     operations = [