Store issue freqs for [release+env]

This adjusts the tsdb storage for release frequencies to use the new GroupRelease which is bound to an environment.

/cc @getsentry/infrastructure

Author: dcramer

src/sentry/event_manager.py

@@ -559,16 +559,41 @@ def save(self, project, raw=False):
 
         if release:
             # TODO(dcramer): would be a bit more ideal to buffer this
-            GroupRelease.objects.create_or_update(
-                project_id=project.id,
-                release_id=release.id,
-                group_id=group.id,
-                environment=environment or '',
-                values={
-                    'last_seen': date,
-                }
+            grouprelease = GroupRelease.get_or_create(
+                group=group,
+                release=release,
+                environment=environment,
+                datetime=date,
+            )
+
+        tsdb.incr_multi([
+            (tsdb.models.group, group.id),
+            (tsdb.models.project, project.id),
+        ], timestamp=event.datetime)
+
+        frequencies = [
+            (tsdb.models.frequent_projects_by_organization, {
+                project.organization_id: {
+                    project.id: 1,
+                },
+            }),
+            (tsdb.models.frequent_issues_by_project, {
+                project.id: {
+                    group.id: 1,
+                },
+            })
+        ]
+        if release:
+            frequencies.append(
+                (tsdb.models.frequent_releases_by_groups, {
+                    group.id: {
+                        grouprelease.id: 1,
+                    },
+                })
             )
 
+        tsdb.record_frequency_multi(frequencies, timestamp=event.datetime)
+
         UserReport.objects.filter(
             project=project, event_id=event_id,
         ).update(group=group)
@@ -761,34 +786,6 @@ def _save_aggregate(self, event, hashes, release, **kwargs):
         else:
             is_sample = can_sample
 
-        tsdb.incr_multi([
-            (tsdb.models.group, group.id),
-            (tsdb.models.project, project.id),
-        ], timestamp=event.datetime)
-
-        frequencies = [
-            (tsdb.models.frequent_projects_by_organization, {
-                project.organization_id: {
-                    project.id: 1,
-                },
-            }),
-            (tsdb.models.frequent_issues_by_project, {
-                project.id: {
-                    group.id: 1,
-                },
-            })
-        ]
-        if release:
-            frequencies.append(
-                (tsdb.models.frequent_releases_by_groups, {
-                    group.id: {
-                        release.id: 1,
-                    },
-                })
-            )
-
-        tsdb.record_frequency_multi(frequencies, timestamp=event.datetime)
-
         return group, is_new, is_regression, is_sample
 
     def _handle_regression(self, group, event, release):

src/sentry/models/grouprelease.py

@@ -2,7 +2,9 @@
 
 from django.db import models
 from django.utils import timezone
+from hashlib import md5
 
+from sentry.utils.cache import cache
 from sentry.db.models import (
     BoundedPositiveIntegerField, Model, sane_repr
 )
@@ -24,3 +26,39 @@ class Meta:
         unique_together = (('group_id', 'release_id', 'environment'),)
 
     __repr__ = sane_repr('group_id', 'release_id')
+
+    @classmethod
+    def get_cache_key(cls, group_id, release_id, environment):
+        return 'grouprelease:1:{}:{}'.format(
+            group_id,
+            md5('{}:{}'.format(release_id, environment)).hexdigest(),
+        )
+
+    @classmethod
+    def get_or_create(cls, group, release, environment, datetime, **kwargs):
+        if not environment:
+            environment = ''
+        cache_key = cls.get_cache_key(group.id, release.id, environment)
+
+        instance = cache.get(cache_key)
+        if instance is None:
+            # TODO(dcramer): if the cache result is -1 we could attempt a
+            # default create here instead of default get
+            instance, created = cls.objects.get_or_create(
+                release_id=release.id,
+                group_id=group.id,
+                environment=environment,
+                defaults={
+                    'project_id': group.project_id,
+                    'first_seen': datetime,
+                    'last_seen': datetime,
+                },
+            )
+            cache.set(cache_key, instance, 3600)
+        else:
+            created = False
+
+        # TODO(dcramer): this would be good to buffer
+        if not created:
+            instance.update(last_seen=datetime)
+        return instance

src/sentry/tsdb/base.py

@@ -62,7 +62,9 @@ class TSDBModel(Enum):
     # number of issues seen for a project, by project
     frequent_issues_by_project = 404
     # number of events seen for a release, by issue
-    frequent_releases_by_groups = 406
+    # frequent_releases_by_groups = 406  # DEPRECATED
+    # number of events seen for a release, by issue
+    frequent_releases_by_groups = 407
 
 
 class BaseTSDB(object):

tests/sentry/models/test_grouprelease.py

@@ -0,0 +1,58 @@
+from __future__ import absolute_import
+
+from datetime import timedelta
+from django.utils import timezone
+
+from sentry.models import GroupRelease, Release
+from sentry.testutils import TestCase
+
+
+class GetOrCreateTest(TestCase):
+    def test_simple(self):
+        project = self.create_project()
+        group = self.create_group(project=project)
+        release = Release.objects.create(version='abc', project=project)
+
+        datetime = timezone.now()
+
+        grouprelease = GroupRelease.get_or_create(
+            group=group,
+            release=release,
+            environment='prod',
+            datetime=datetime,
+        )
+
+        assert grouprelease.project_id == project.id
+        assert grouprelease.group_id == group.id
+        assert grouprelease.release_id == release.id
+        assert grouprelease.environment == 'prod'
+        assert grouprelease.first_seen == datetime
+        assert grouprelease.last_seen == datetime
+
+        datetime_new = timezone.now() + timedelta(days=1)
+
+        grouprelease = GroupRelease.get_or_create(
+            group=group,
+            release=release,
+            environment='prod',
+            datetime=datetime_new,
+        )
+
+        assert grouprelease.first_seen == datetime
+        assert grouprelease.last_seen == datetime_new
+
+    def test_empty_env(self):
+        project = self.create_project()
+        group = self.create_group(project=project)
+        release = Release.objects.create(version='abc', project=project)
+
+        datetime = timezone.now()
+
+        grouprelease = GroupRelease.get_or_create(
+            group=group,
+            release=release,
+            environment=None,
+            datetime=datetime,
+        )
+
+        assert grouprelease.environment == ''