service-capacity-modeling 0.3.57__tar.gz → 0.3.59__tar.gz

{service_capacity_modeling-0.3.57 → service_capacity_modeling-0.3.59}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: service-capacity-modeling
-Version: 0.3.57
+Version: 0.3.59
 Summary: Contains utilities for modeling capacity for pluggable workloads
 Author: Joseph Lynch
 Author-email: josephl@netflix.com

{service_capacity_modeling-0.3.57 → service_capacity_modeling-0.3.59}/service_capacity_modeling/models/org/netflix/cassandra.py

@@ -4,6 +4,7 @@ from typing import Any
 from typing import Callable
 from typing import Dict
 from typing import Optional
+from typing import Set
 
 from pydantic import BaseModel
 from pydantic import Field
@@ -47,6 +48,9 @@ from service_capacity_modeling.stats import dist_for_interval
 logger = logging.getLogger(__name__)
 
 BACKGROUND_BUFFER = "background"
+CRITICAL_TIERS: Set[int] = {0, 1}
+# cluster size aka nodes per ASG
+CRITICAL_TIER_MIN_CLUSTER_SIZE = 2
 
 
 def _write_buffer_gib_zone(
@@ -362,10 +366,7 @@ def _estimate_cassandra_cluster_zonal(  # pylint: disable=too-many-positional-ar
     # Cassandra clusters should aim to be at least 2 nodes per zone to start
     # out with for tier 0 or tier 1. This gives us more room to "up-color"]
     # clusters.
-    min_count = 0
-    if desires.service_tier <= 1:
-        min_count = 2
-
+    min_count = 2 if desires.service_tier in CRITICAL_TIERS else 0
     base_mem = _get_base_memory(desires)
 
     heap_fn = _cass_heap_for_write_buffer(
@@ -583,6 +584,33 @@ class NflxCassandraArguments(BaseModel):
 
 
 class NflxCassandraCapacityModel(CapacityModel):
+    @staticmethod
+    def get_required_cluster_size(tier, extra_model_arguments):
+        required_cluster_size: Optional[int] = (
+            math.ceil(extra_model_arguments["required_cluster_size"])
+            if "required_cluster_size" in extra_model_arguments
+            else None
+        )
+
+        if tier not in CRITICAL_TIERS or required_cluster_size is None:
+            return required_cluster_size
+
+        # If the upstream explicitly set a cluster size, make sure it is
+        # at least CRITICAL_TIER_MIN_CLUSTER_SIZE. We cannot do a max
+        # of the two because the horizontal scaling is disabled
+        if required_cluster_size < CRITICAL_TIER_MIN_CLUSTER_SIZE:
+            raise ValueError(
+                f"Required cluster size must be at least "
+                f"{CRITICAL_TIER_MIN_CLUSTER_SIZE=} when "
+                f"service tier({tier}) is a "
+                f"critical tier({CRITICAL_TIERS})."
+                f"If it is an existing cluster, horizontally "
+                f"scale the cluster to be >= "
+                f"{CRITICAL_TIER_MIN_CLUSTER_SIZE}"
+            )
+
+        return required_cluster_size
+
     @staticmethod
     def capacity_plan(
         instance: Instance,
@@ -601,11 +629,12 @@ class NflxCassandraCapacityModel(CapacityModel):
         require_attached_disks: bool = extra_model_arguments.get(
             "require_attached_disks", False
         )
-        required_cluster_size: Optional[int] = (
-            math.ceil(extra_model_arguments["required_cluster_size"])
-            if "required_cluster_size" in extra_model_arguments
-            else None
+        required_cluster_size: Optional[
+            int
+        ] = NflxCassandraCapacityModel.get_required_cluster_size(
+            desires.service_tier, extra_model_arguments
         )
+
         max_rps_to_disk: int = extra_model_arguments.get("max_rps_to_disk", 500)
         max_regional_size: int = extra_model_arguments.get("max_regional_size", 192)
         max_local_disk_gib: int = extra_model_arguments.get("max_local_disk_gib", 5120)

{service_capacity_modeling-0.3.57 → service_capacity_modeling-0.3.59}/service_capacity_modeling/tools/generate_missing.py

@@ -40,7 +40,7 @@ def build_command(family: str, params: Dict[str, Any], output_path: Path) -> lis
     return cmd
 
 
-def main(debug: bool = True, execute: bool = False):
+def main(debug: bool = True, execute: bool = False, force: bool = False):
     expected_path = (
         Path(__file__).resolve().parent.parent / "hardware/profiles/shapes/aws"
     )
@@ -51,7 +51,7 @@ def main(debug: bool = True, execute: bool = False):
     missing_families = {
         family: params
         for family, params in INSTANCE_TYPES.items()
-        if not (expected_path / f"auto_{family}.json").exists()
+        if force or not (expected_path / f"auto_{family}.json").exists()
     }
 
     if not missing_families:
@@ -87,5 +87,6 @@ if __name__ == "__main__":
     # Parse arguments to determine whether to execute commands
     execute_mode = "--execute" in sys.argv
     debug_mode = "--debug" in sys.argv or not execute_mode
+    force_mode = "--force" in sys.argv
 
-    main(debug=debug_mode, execute=execute_mode)
+    main(debug=debug_mode, execute=execute_mode, force=force_mode)

{service_capacity_modeling-0.3.57 → service_capacity_modeling-0.3.59}/service_capacity_modeling.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: service-capacity-modeling
-Version: 0.3.57
+Version: 0.3.59
 Summary: Contains utilities for modeling capacity for pluggable workloads
 Author: Joseph Lynch
 Author-email: josephl@netflix.com