feat: add option to set cpu limits (#1384)

Author: olevski

components/renku_data_services/crc/api.spec.yaml

@@ -1543,6 +1543,8 @@ components:
           $ref: "#/components/schemas/Ulid"
         platform:
           $ref: "#/components/schemas/RuntimePlatform"
+        cpu_limit_factor:
+          $ref: "#/components/schemas/CpuLimitFactor"
       required: ["classes", "name", "public", "default"]
       example:
         quota:
@@ -1595,6 +1597,10 @@ components:
           $ref: "#/components/schemas/Ulid"
         platform:
           $ref: "#/components/schemas/RuntimePlatform"
+        cpu_limit_factor:
+          oneOf:
+            - $ref: "#/components/schemas/CpuLimitFactor"
+            - $ref: "#/components/schemas/RemoveCpuLimitFactor"
       example:
         quota:
           cpu: 50
@@ -1631,6 +1637,10 @@ components:
           $ref: "#/components/schemas/Ulid"
         platform:
           $ref: "#/components/schemas/RuntimePlatform"
+        cpu_limit_factor:
+          oneOf:
+            - $ref: "#/components/schemas/CpuLimitFactor"
+            - $ref: "#/components/schemas/RemoveCpuLimitFactor"
       required: ["classes", "name", "public", "default", "platform"]
       example:
         quota:
@@ -1690,6 +1700,8 @@ components:
           required: ["id"]
         platform:
           $ref: "#/components/schemas/RuntimePlatform"
+        cpu_limit_factor:
+          $ref: "#/components/schemas/CpuLimitFactor"
       required: ["classes", "name", "id", "public", "default", "platform"]
       example:
         quota:
@@ -1748,6 +1760,8 @@ components:
           $ref: "#/components/schemas/RuntimePlatform"
         credits_used:
           $ref: "#/components/schemas/CreditsUsed"
+        cpu_limit_factor:
+          $ref: "#/components/schemas/CpuLimitFactor"
       required: ["classes", "name", "id", "public", "default", "platform"]
       example:
         quota:
@@ -2159,7 +2173,7 @@ components:
       properties:
         kind:
           type: string
-          enum: [ "firecrest" ]
+          enum: ["firecrest"]
           description: Kind of remote resource pool
           example: "firecrest"
         provider_id:
@@ -2200,7 +2214,7 @@ components:
       properties:
         kind:
           type: string
-          enum: [ "runai" ]
+          enum: ["runai"]
           description: Kind of remote resource pool
           example: "runai"
         base_url:
@@ -2235,7 +2249,7 @@ components:
       properties:
         kind:
           type: string
-          enum: [ "firecrest" ]
+          enum: ["firecrest"]
           description: Kind of remote resource pool
           example: "firecrest"
         provider_id:
@@ -2254,7 +2268,7 @@ components:
       properties:
         kind:
           type: string
-          enum: [ "runai" ]
+          enum: ["runai"]
           description: Kind of remote resource pool
           example: "runai"
         base_url:
@@ -2339,7 +2353,7 @@ components:
       description: A list of k8s labels used for tolerations and/or node affinity
       items:
         $ref: "#/components/schemas/NodeAffinity"
-      example: [{"key": "test-label-1", "required_during_scheduling": false}]
+      example: [{ "key": "test-label-1", "required_during_scheduling": false }]
       uniqueItems: true
       default: []
       minItems: 0
@@ -2348,7 +2362,7 @@ components:
       description: A list of k8s labels used for tolerations and/or node affinity
       items:
         $ref: "#/components/schemas/NodeAffinity"
-      example: [{"key": "test-label-1", "required_during_scheduling": false}]
+      example: [{ "key": "test-label-1", "required_during_scheduling": false }]
       uniqueItems: true
       default: []
       minItems: 0
@@ -2385,11 +2399,11 @@ components:
       type: string
       minLength: 26
       maxLength: 26
-      pattern: "^[0-7][0-9A-HJKMNP-TV-Z]{25}$"  # This is case-insensitive
+      pattern: "^[0-7][0-9A-HJKMNP-TV-Z]{25}$" # This is case-insensitive
     Protocol:
       description: Allowed Protocol strings
       type: string
-      enum: [ "http", "https" ]
+      enum: ["http", "https"]
     Port:
       type: integer
       minimum: 0
@@ -2411,6 +2425,18 @@ components:
     IngressAnnotations:
       type: object
       additionalProperties: true
+    CpuLimitFactor:
+      type: number
+      format: float
+      minimum: 1
+      maximum: 100
+      description: The value used to multiply cpu requests to get the limit. If unset there will be no cpu limits.
+    RemoveCpuLimitFactor:
+      type: number
+      minimum: 0
+      maximum: 0
+      description: Indicates that the cpu limits should not be set and if a limit is set it should be removed
+      format: float
   responses:
     Error:
       description: The schema for all 4xx and 5xx responses

components/renku_data_services/crc/apispec.py

@@ -1,6 +1,6 @@
 # generated by datamodel-codegen:
 #   filename:  api.spec.yaml
-#   timestamp: 2026-06-29T12:51:11+00:00
+#   timestamp: 2026-07-16T20:59:27+00:00
 
 from __future__ import annotations
 
@@ -219,6 +219,24 @@ class IngressAnnotations(BaseAPISpec):
     )
 
 
+class CpuLimitFactor(RootModel[float]):
+    root: float = Field(
+        ...,
+        description="The value used to multiply cpu requests to get the limit. If unset there will be no cpu limits.",
+        ge=1.0,
+        le=100.0,
+    )
+
+
+class RemoveCpuLimitFactor(RootModel[float]):
+    root: float = Field(
+        ...,
+        description="Indicates that the cpu limits should not be set and if a limit is set it should be removed",
+        ge=0.0,
+        le=0.0,
+    )
+
+
 class ResourcePoolsParams(BaseAPISpec):
     model_config = ConfigDict(
         extra="forbid",
@@ -984,6 +1002,7 @@ class ResourcePoolPatch(BaseAPISpec):
         pattern="^[0-7][0-9A-HJKMNP-TV-Z]{25}$",
     )
     platform: RuntimePlatform | None = None
+    cpu_limit_factor: CpuLimitFactor | RemoveCpuLimitFactor | None = None
 
 
 class ResourcePool(BaseAPISpec):
@@ -1038,6 +1057,12 @@ class ResourcePool(BaseAPISpec):
         pattern="^[0-7][0-9A-HJKMNP-TV-Z]{25}$",
     )
     platform: RuntimePlatform | None = None
+    cpu_limit_factor: float | None = Field(
+        None,
+        description="The value used to multiply cpu requests to get the limit. If unset there will be no cpu limits.",
+        ge=1.0,
+        le=100.0,
+    )
 
 
 class ResourcePoolPut(BaseAPISpec):
@@ -1120,6 +1145,7 @@ class ResourcePoolPut(BaseAPISpec):
         pattern="^[0-7][0-9A-HJKMNP-TV-Z]{25}$",
     )
     platform: RuntimePlatform
+    cpu_limit_factor: CpuLimitFactor | RemoveCpuLimitFactor | None = None
 
 
 class ResourcePoolWithId(BaseAPISpec):
@@ -1174,6 +1200,12 @@ class ResourcePoolWithId(BaseAPISpec):
     )
     cluster: Cluster1 | None = None
     platform: RuntimePlatform
+    cpu_limit_factor: float | None = Field(
+        None,
+        description="The value used to multiply cpu requests to get the limit. If unset there will be no cpu limits.",
+        ge=1.0,
+        le=100.0,
+    )
 
 
 class ResourcePoolWithIdFiltered(BaseAPISpec):
@@ -1240,6 +1272,12 @@ class ResourcePoolWithIdFiltered(BaseAPISpec):
         examples=[300],
         ge=0,
     )
+    cpu_limit_factor: float | None = Field(
+        None,
+        description="The value used to multiply cpu requests to get the limit. If unset there will be no cpu limits.",
+        ge=1.0,
+        le=100.0,
+    )
 
 
 class ResourcePoolsWithId(RootModel[list[ResourcePoolWithId]]):

components/renku_data_services/crc/core.py

@@ -235,6 +235,16 @@ def validate_resource_pool_put_or_patch(
     hibernation_warning_period = validate_hibernation_warning_period(
         body.hibernation_threshold, body.hibernation_warning_period
     )
+
+    cpu_limit_factorrr: None | ResetType | float = None
+    match method, body.cpu_limit_factor:
+        case _, apispec.RemoveCpuLimitFactor():
+            cpu_limit_factorrr = RESET
+        case "PUT", None:
+            cpu_limit_factorrr = RESET
+        case _, apispec.CpuLimitFactor():
+            cpu_limit_factorrr = body.cpu_limit_factor.root
+
     return models.ResourcePoolPatch(
         name=body.name,
         classes=classes,
@@ -247,6 +257,7 @@ def validate_resource_pool_put_or_patch(
         remote=remote,
         cluster_id=ULID.from_str(body.cluster_id) if body.cluster_id else None,
         platform=platform,
+        cpu_limit_factor=cpu_limit_factorrr,
     )
 
 

components/renku_data_services/crc/db.py

@@ -622,6 +622,10 @@ async def update_resource_pool(
                 rp.hibernation_warning_period = update.hibernation_warning_period
             if update.platform is not None:
                 rp.platform = update.platform
+            if update.cpu_limit_factor == 0 or update.cpu_limit_factor is RESET:
+                rp.cpu_limit_factor = None
+            elif update.cpu_limit_factor is not None:
+                rp.cpu_limit_factor = update.cpu_limit_factor
 
             match (update.cluster_id, rp.cluster_id):
                 case ResetType.Reset, x if x is not None:

components/renku_data_services/crc/models.py

@@ -333,6 +333,10 @@ class UnsavedResourcePool:
     remote: RemoteConfigurationFirecrest | RemoteConfigurationRunai | None = None
     cluster_id: ClusterId | None = None
     platform: RuntimePlatform
+    cpu_limit_factor: float | None = None
+    """Used to assign cpu limits based on the cpu value in the resource classes in the pool.
+    If the value is zero or unset then cpu limits are not set.
+    """
 
 
 @dataclass(frozen=True, eq=True, kw_only=True)
@@ -352,6 +356,10 @@ class ResourcePool:
     cluster: SavedClusterSettings | None = None
     platform: RuntimePlatform
     credits_used: int | None = None
+    cpu_limit_factor: float | None = None
+    """Used to assign cpu limits based on the cpu value in the resource classes in the pool.
+    If the value is zero or unset then cpu limits are not set.
+    """
 
     def get_resource_class(self, resource_class_id: int) -> ResourceClass | None:
         """Find a specific resource class in the resource pool by the resource class id."""
@@ -389,6 +397,7 @@ class ResourcePoolPatch:
     remote: RemoteConfigurationPatch | None = None
     cluster_id: ClusterId | ResetType | None = None
     platform: RuntimePlatform | None = None
+    cpu_limit_factor: float | None | ResetType = None
 
 
 class RemoteConfigurationKind(StrEnum):

components/renku_data_services/crc/orm.py

@@ -10,6 +10,7 @@
     CheckConstraint,
     Column,
     Enum,
+    Float,
     Identity,
     Integer,
     MetaData,
@@ -254,6 +255,12 @@ class ResourcePoolORM(BaseORM):
     """Resource pool specifies a set of resource classes, users that can access them and a quota."""
 
     __tablename__ = "resource_pools"
+    __table_args__ = (
+        CheckConstraint(
+            "cpu_limit_factor IS NULL OR cpu_limit_factor >= 1.0",
+            name="chk_cpu_limit_factor_gte_1",
+        ),
+    )
     name: Mapped[str] = mapped_column(String(40), index=True)
     quota: Mapped[Optional[str]] = mapped_column(String(63), index=True, default=None)
     users: Mapped[list[UserORM]] = relationship(
@@ -298,6 +305,11 @@ class ResourcePoolORM(BaseORM):
         Enum(models.RuntimePlatform, name="build_platform"), default=None, server_default=literal("linux_amd64")
     )
 
+    cpu_limit_factor: Mapped[Optional[float]] = mapped_column("cpu_limit_factor", Float, default=None, nullable=True)
+    """Used to assign cpu limits based on the cpu value in the resource classes in the pool.
+    If the value is zero or unset then cpu limits are not set.
+    """
+
     @classmethod
     def from_unsaved_model(
         cls,
@@ -329,6 +341,7 @@ def from_unsaved_model(
             remote_json=remote_json,
             cluster_id=cluster.id if cluster else None,
             platform=new_resource_pool.platform,
+            cpu_limit_factor=new_resource_pool.cpu_limit_factor,
         )
 
     def dump(
@@ -366,6 +379,7 @@ def dump(
             cluster=cluster,
             platform=self.platform,
             credits_used=credits_used.value if credits_used else None,
+            cpu_limit_factor=self.cpu_limit_factor,
         )
 
     def _dump_remote(self) -> models.RemoteConfigurationFirecrest | models.RemoteConfigurationRunai | None:

components/renku_data_services/migrations/versions/5cb5a8d5663b_add_cpu_limit_factor.py

@@ -0,0 +1,35 @@
+"""add cpu limit factor
+
+Revision ID: 5cb5a8d5663b
+Revises: 0a6cee40fe0d
+Create Date: 2026-07-13 13:14:07.765045
+
+"""
+
+import sqlalchemy as sa
+from alembic import op
+
+# revision identifiers, used by Alembic.
+revision = "5cb5a8d5663b"
+down_revision = "0a6cee40fe0d"
+branch_labels = None
+depends_on = None
+
+
+def upgrade() -> None:
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.add_column("resource_pools", sa.Column("cpu_limit_factor", sa.Float(), nullable=True), schema="resource_pools")
+    # ### end Alembic commands ###
+    op.create_check_constraint(
+        "chk_cpu_limit_factor_gte_1",
+        "resource_pools",
+        "cpu_limit_factor IS NULL OR cpu_limit_factor >= 1.0",
+        schema="resource_pools",
+    )
+
+
+def downgrade() -> None:
+    op.drop_constraint("chk_cpu_limit_factor_gte_1", "resource_pools", schema="resource_pools", type_="check")
+    # ### commands auto generated by Alembic - please adjust! ###
+    op.drop_column("resource_pools", "cpu_limit_factor", schema="resource_pools")
+    # ### end Alembic commands ###