add option to filter amount of orgs

Author: Andrii Soldatenko

src/sentry/dynamic_sampling/prioritise_projects.py

@@ -17,6 +17,7 @@
     Request,
 )
 
+from sentry import options
 from sentry.dynamic_sampling.rules.utils import OrganizationId, ProjectId
 from sentry.sentry_metrics.indexer.strings import TRANSACTION_METRICS_NAMES
 from sentry.snuba.dataset import Dataset, EntityKey
@@ -35,6 +36,7 @@ def fetch_projects_with_total_volumes() -> Mapping[OrganizationId, Sequence[Tupl
     aggregated_projects = defaultdict(list)
     start_time = time.time()
     offset = 0
+    sample_rate = int(options.get("dynamic-sampling.prioritise_projects.sample_rate") * 100)
     while (time.time() - start_time) < MAX_SECONDS:
         query = (
             Query(
@@ -46,6 +48,7 @@ def fetch_projects_with_total_volumes() -> Mapping[OrganizationId, Sequence[Tupl
                 ],
                 groupby=[Column("org_id"), Column("project_id")],
                 where=[
+                    Condition(Function("modulo", [Column("org_id"), 100]), Op.LT, sample_rate),
                     Condition(Column("timestamp"), Op.GTE, datetime.utcnow() - timedelta(hours=6)),
                     Condition(Column("timestamp"), Op.LT, datetime.utcnow()),
                     Condition(

src/sentry/options/defaults.py

@@ -606,6 +606,7 @@
 # System-wide options that observes latest releases on transactions and caches these values to be used later in
 # project config computation. This is temporary option to monitor the performance of this feature.
 register("dynamic-sampling:boost-latest-release", default=False)
+register("dynamic-sampling.prioritise_projects.sample_rate", default=0.0)
 
 # Killswitch for deriving code mappings
 register("post_process.derive-code-mappings", default=True)

tests/sentry/dynamic_sampling/test_prioritise_projects.py

@@ -30,6 +30,8 @@ def test_simple_one_org_one_project(self):
             project_id=p1.id,
             org_id=org1.id,
         )
-
-        results = fetch_projects_with_total_volumes()
+        with self.settings(
+            SENTRY_OPTIONS={"dynamic-sampling.prioritise_projects.sample_rate": 1.0}
+        ):
+            results = fetch_projects_with_total_volumes()
         assert results[org1.id] == [(p1.id, 1.0)]

tests/sentry/dynamic_sampling/test_tasks.py

@@ -58,9 +58,12 @@ def test_prioritise_projects_simple(self, get_blended_sample_rate):
         proj_c = self.create_project_and_add_metrics("c", 3, test_org)
         proj_d = self.create_project_and_add_metrics("d", 1, test_org)
 
-        with self.feature({"organizations:ds-prioritise-by-project-bias": True}):
-            with self.tasks():
-                prioritise_projects()
+        with self.settings(
+            SENTRY_OPTIONS={"dynamic-sampling.prioritise_projects.sample_rate": 1.0}
+        ):
+            with self.feature({"organizations:ds-prioritise-by-project-bias": True}):
+                with self.tasks():
+                    prioritise_projects()
 
         # we expect only uniform rule
         assert generate_rules(proj_a)[0]["samplingValue"] == {