runnable 0.17.1__py3-none-any.whl → 0.18.0__py3-none-any.whl

extensions/job_executor/k8s.py

@@ -0,0 +1,362 @@
+import logging
+import shlex
+from enum import Enum
+from typing import Annotated, List, Optional
+
+from kubernetes import client
+from kubernetes import config as k8s_config
+from pydantic import BaseModel, ConfigDict, Field, PlainSerializer, PrivateAttr
+from pydantic.alias_generators import to_camel
+
+from extensions.job_executor import GenericJobExecutor
+from runnable import console, defaults, utils
+from runnable.datastore import DataCatalog
+from runnable.tasks import BaseTaskType
+
+logger = logging.getLogger(defaults.NAME)
+
+
+class Operator(str, Enum):
+    NOT_IN = "NotIn"
+    EXISTS = "Exists"
+    DOES_NOT_EXIST = "DoesNotExist"
+
+
+class RestartPolicy(str, Enum):
+    ALWAYS = "Always"
+    ON_FAILURE = "OnFailure"
+    NEVER = "Never"
+
+
+class ImagePullPolicy(str, Enum):
+    ALWAYS = "Always"
+    IF_NOT_PRESENT = "IfNotPresent"
+    NEVER = "Never"
+
+
+class LabelSelectorRequirement(BaseModel):
+    key: str
+    operator: Operator
+    values: list[str]
+
+
+class LabelSelector(BaseModel):
+    match_expressions: list[LabelSelectorRequirement]
+    match_labels: dict[str, str]
+
+
+class ObjectMetaData(BaseModel):
+    generate_name: Optional[str]
+    annotations: Optional[dict[str, str]]
+    namespace: Optional[str] = "default"
+
+
+class EnvVar(BaseModel):
+    name: str
+    value: str
+
+
+class Request(BaseModel):
+    """
+    The default requests
+    """
+
+    memory: str = "1Gi"
+    cpu: str = "250m"
+
+
+VendorGPU = Annotated[
+    Optional[int],
+    PlainSerializer(lambda x: str(x), return_type=str, when_used="unless-none"),
+]
+
+
+class Limit(Request):
+    """
+    The default limits
+    """
+
+    gpu: VendorGPU = Field(default=None, serialization_alias="nvidia.com/gpu")
+
+
+class Resources(BaseModel):
+    limits: Limit = Limit()
+    requests: Request = Request()
+
+
+class VolumeMount(BaseModel):
+    name: str
+    mount_path: str
+
+
+class Container(BaseModel):
+    image: str
+    env: list[EnvVar] = Field(default_factory=list)
+    image_pull_policy: ImagePullPolicy = ImagePullPolicy.NEVER
+    resources: Resources = Resources()
+    volume_mounts: Optional[list[VolumeMount]] = Field(default_factory=lambda: [])
+
+
+class HostPath(BaseModel):
+    path: str
+
+
+class Volume(BaseModel):
+    name: str
+    host_path: HostPath
+
+
+class TemplateSpec(BaseModel):
+    active_deadline_seconds: int = Field(default=60 * 60 * 2)  # 2 hours
+    node_selector: Optional[dict[str, str]] = None
+    tolerations: Optional[list[dict[str, str]]] = None
+    volumes: Optional[list[Volume]] = Field(default_factory=lambda: [])
+    service_account_name: Optional[str] = "default"
+    restart_policy: RestartPolicy = RestartPolicy.NEVER
+    container: Container
+
+
+class Template(BaseModel):
+    spec: TemplateSpec
+    metadata: Optional[ObjectMetaData] = None
+
+
+class Spec(BaseModel):
+    active_deadline_seconds: Optional[int] = Field(default=60 * 60 * 2)  # 2 hours
+    backoff_limit: int = 6
+    selector: Optional[LabelSelector] = None
+    template: Template
+    ttl_seconds_after_finished: Optional[int] = Field(default=60 * 60 * 24)  # 24 hours
+
+
+class K8sJobExecutor(GenericJobExecutor):
+    service_name: str = "k8s-job"
+    config_path: Optional[str] = None
+    job_spec: Spec
+    mock: bool = False
+
+    # The location the mount of .run_log_store is mounted to in minikube
+    # ensure that minikube mount $HOME/workspace/runnable/.run_log_store:/volume/run_logs is executed first
+    # $HOME/workspace/runnable/.catalog:/volume/catalog
+    # Ensure that the docker build is done with eval $(minikube docker-env)
+    mini_k8s_run_log_location: str = Field(default="/volume/run_logs/")
+    mini_k8s_catalog_location: str = Field(default="/volume/catalog/")
+
+    _is_local: bool = PrivateAttr(default=False)
+
+    _container_log_location: str = PrivateAttr(default="/tmp/run_logs/")
+    _container_catalog_location: str = PrivateAttr(default="/tmp/catalog/")
+    _container_secrets_location: str = PrivateAttr(default="/tmp/dotenv")
+
+    _volumes: list[Volume] = []
+    _volume_mounts: list[VolumeMount] = []
+
+    model_config = ConfigDict(
+        alias_generator=to_camel,
+        populate_by_name=True,
+        from_attributes=True,
+    )
+
+    def submit_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        This method gets invoked by the CLI.
+        """
+        self._set_up_run_log()
+
+        # Call the container job
+        job_log = self._context.run_log_store.create_job_log()
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )
+        # create volumes and volume mounts for the job
+        self._create_volumes()
+        self.submit_k8s_job(job)
+
+    def execute_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        Focusses on execution of the job.
+        """
+        logger.info("Trying to execute job")
+        self._use_volumes()
+
+        job_log = self._context.run_log_store.get_job_log(run_id=self._context.run_id)
+
+        attempt_log = job.execute_command(
+            attempt_number=self.step_attempt_number,
+            mock=self.mock,
+        )
+
+        job_log.status = attempt_log.status
+        job_log.attempts.append(attempt_log)
+
+        data_catalogs_put: Optional[List[DataCatalog]] = self._sync_catalog(
+            catalog_settings=catalog_settings
+        )
+        logger.debug(f"data_catalogs_put: {data_catalogs_put}")
+
+        job_log.add_data_catalogs(data_catalogs_put or [])
+
+        console.print("Summary of job")
+        console.print(job_log.get_summary())
+
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )
+
+    @property
+    def _client(self):
+        if self.config_path:
+            k8s_config.load_kube_config(config_file=self.config_path)
+        else:
+            # https://github.com/kubernetes-client/python/blob/master/kubernetes/base/config/__init__.py
+            k8s_config.load_config()
+        return client
+
+    def submit_k8s_job(self, task: BaseTaskType):
+        if self.job_spec.template.spec.container.volume_mounts:
+            self._volume_mounts += self.job_spec.template.spec.container.volume_mounts
+
+        container_volume_mounts = [
+            self._client.V1VolumeMount(**vol.model_dump())
+            for vol in self._volume_mounts
+        ]
+
+        command = utils.get_job_execution_command()
+
+        container_env = [
+            self._client.V1EnvVar(**env.model_dump(by_alias=True))
+            for env in self.job_spec.template.spec.container.env
+        ]
+
+        base_container = self._client.V1Container(
+            command=shlex.split(command),
+            env=container_env,
+            name="default",
+            volume_mounts=container_volume_mounts,
+            **self.job_spec.template.spec.container.model_dump(
+                exclude_none=True, exclude={"volume_mounts", "command", "env"}
+            ),
+        )
+
+        if self.job_spec.template.spec.volumes:
+            self._volumes += self.job_spec.template.spec.volumes
+
+        spec_volumes = [
+            self._client.V1Volume(**vol.model_dump(by_alias=True))
+            for vol in self._volumes
+        ]
+
+        tolerations = None
+        if self.job_spec.template.spec.tolerations:
+            tolerations = [
+                self._client.V1Toleration(**toleration)
+                for toleration in self.job_spec.template.spec.tolerations
+            ]
+
+        pod_spec = self._client.V1PodSpec(
+            containers=[base_container],
+            # volumes=[vol.model_dump(by_alias=True) for vol in self._volumes],
+            volumes=spec_volumes,
+            tolerations=tolerations,
+            **self.job_spec.template.spec.model_dump(
+                exclude_none=True, exclude={"container", "volumes", "tolerations"}
+            ),
+        )
+
+        pod_template_metadata = None
+        if self.job_spec.template.metadata:
+            pod_template_metadata = self._client.V1ObjectMeta(
+                **self.job_spec.template.metadata.model_dump(exclude_none=True)
+            )
+
+        pod_template = self._client.V1PodTemplateSpec(
+            spec=pod_spec,
+            metadata=pod_template_metadata,
+        )
+
+        job_spec = client.V1JobSpec(
+            template=pod_template,
+            **self.job_spec.model_dump(exclude_none=True, exclude={"template"}),
+        )
+
+        job = client.V1Job(
+            api_version="batch/v1",
+            kind="Job",
+            metadata=client.V1ObjectMeta(name=self._context.run_id),
+            spec=job_spec,
+        )
+
+        logger.info(f"Submitting job: {job.__dict__}")
+
+        try:
+            k8s_batch = self._client.BatchV1Api()
+            response = k8s_batch.create_namespaced_job(
+                body=job,
+                namespace="default",
+                _preload_content=False,
+                pretty=True,
+            )
+            logger.debug(f"Kubernetes job response: {response}")
+        except Exception as e:
+            logger.exception(e)
+            print(e)
+            raise
+
+    def _create_volumes(self):
+        match self._context.run_log_store.service_name:
+            case "file-system":
+                self._volumes.append(
+                    # When you do: # minikube mount $HOME:/tmp/run_logs
+                    # This .run_log_store is mounted to /tmp/run_logs of minikube
+                    # You then are creating a volume that is mounted to /tmp/run_logs in the container
+                    # You are then referring to it.
+                    # https://stackoverflow.com/questions/57411456/minikube-mounted-host-folders-are-not-working
+                    Volume(
+                        name="run-logs",
+                        host_path=HostPath(path=self.mini_k8s_run_log_location),
+                    )
+                )
+                self._volume_mounts.append(
+                    VolumeMount(
+                        name="run-logs", mount_path=self._container_log_location
+                    )
+                )
+            case "chunked-fs":
+                self._volumes.append(
+                    Volume(
+                        name="run-logs",
+                        host_path=HostPath(path=self.mini_k8s_run_log_location),
+                    )
+                )
+                self._volume_mounts.append(
+                    VolumeMount(
+                        name="run-logs", mount_path=self._container_log_location
+                    )
+                )
+
+        match self._context.catalog_handler.service_name:
+            case "file-system":
+                self._volumes.append(
+                    Volume(
+                        name="catalog",
+                        host_path=HostPath(path=self.mini_k8s_catalog_location),
+                    )
+                )
+                self._volume_mounts.append(
+                    VolumeMount(
+                        name="catalog", mount_path=self._container_catalog_location
+                    )
+                )
+
+    def _use_volumes(self):
+        match self._context.run_log_store.service_name:
+            case "file-system":
+                self._context.run_log_store.log_folder = self._container_log_location
+            case "chunked-fs":
+                self._context.run_log_store.log_folder = self._container_log_location
+
+        match self._context.catalog_handler.service_name:
+            case "file-system":
+                self._context.catalog_handler.catalog_location = (
+                    self._container_catalog_location
+                )

extensions/job_executor/k8s_job_spec.yaml

@@ -0,0 +1,37 @@
+#Follow this as a template https://github.com/kubernetes-client/python/blob/master/kubernetes/docs/V1JobSpec.md
+
+jobSpec:
+  activeDeadlineSeconds: Optional[int]
+  selector: Optional[LabelSelector]
+  ttlSecondsAfterFinished: Optional[int]
+  template:
+    metadata:
+      annotations: Optional[Dict[str, str]]
+      generate_name: Optional[str] = run_id
+      namespace: Optional[str] = "default"
+    spec:
+      activeDeadlineSeconds: Optional[int]
+      nodeSelector: Optional[Dict[str, str]]
+      tolerations: Optional[List[Toleration]]
+      volumes: Optional[List[str]]
+      serviceAccountName: Optional[str]
+      restartPolicy: Optional[str] = Choose from [Always, OnFailure, Never]
+      container:
+        command: List[str]
+        env:
+            - name: str
+              value: str
+        image: str
+        imagePullPolicy: Optional[str] = choose from [Always, Never, IfNotPresent]
+        resources:
+          limits:
+            cpu: str
+            memory: str
+            gpu: str
+          requests:
+            cpu: str
+            memory: str
+            gpu: str
+        volumeMounts:
+          - name: str
+            mountPath: str

extensions/job_executor/local.py

@@ -0,0 +1,61 @@
+import logging
+from typing import List, Optional
+
+from extensions.job_executor import GenericJobExecutor
+from runnable import console, defaults
+from runnable.datastore import DataCatalog
+from runnable.tasks import BaseTaskType
+
+logger = logging.getLogger(defaults.LOGGER_NAME)
+
+
+class LocalJobExecutor(GenericJobExecutor):
+    """
+    The LocalJobExecutor is a job executor that runs the job locally.
+    """
+
+    service_name: str = "local"
+    mock: bool = False
+
+    def submit_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        This method gets invoked by the CLI.
+        """
+        self._set_up_run_log()
+
+        job_log = self._context.run_log_store.create_job_log()
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )
+
+        self.execute_job(job, catalog_settings=catalog_settings)
+
+    def execute_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        Focusses on execution of the job.
+        """
+        logger.info("Trying to execute job")
+
+        job_log = self._context.run_log_store.get_job_log(run_id=self._context.run_id)
+
+        attempt_log = job.execute_command(
+            attempt_number=self.step_attempt_number,
+            mock=self.mock,
+        )
+
+        job_log.status = attempt_log.status
+        job_log.attempts.append(attempt_log)
+
+        data_catalogs_put: Optional[List[DataCatalog]] = self._sync_catalog(
+            catalog_settings=catalog_settings
+        )
+        logger.debug(f"data_catalogs_put: {data_catalogs_put}")
+
+        job_log.add_data_catalogs(data_catalogs_put or [])
+
+        console.print("Summary of job")
+        console.print(job_log.get_summary())
+
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )

extensions/job_executor/local_container.py

@@ -0,0 +1,192 @@
+import logging
+from pathlib import Path
+from typing import Dict, List, Optional
+
+from pydantic import Field
+
+from extensions.job_executor import GenericJobExecutor
+from runnable import console, defaults, utils
+from runnable.datastore import DataCatalog
+from runnable.tasks import BaseTaskType
+
+logger = logging.getLogger(defaults.LOGGER_NAME)
+
+
+class LocalContainerJobExecutor(GenericJobExecutor):
+    """
+    The LocalJobExecutor is a job executor that runs the job locally.
+    """
+
+    service_name: str = "local-container"
+    docker_image: str
+    mock: bool = False
+    auto_remove_container: bool = True
+    environment: Dict[str, str] = Field(default_factory=dict)
+
+    _is_local: bool = False
+
+    _container_log_location = "/tmp/run_logs/"
+    _container_catalog_location = "/tmp/catalog/"
+    _container_secrets_location = "/tmp/dotenv"
+    _volumes: Dict[str, Dict[str, str]] = {}
+
+    def submit_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        This method gets invoked by the CLI.
+        """
+        self._set_up_run_log()
+        self._mount_volumes()
+
+        # Call the container job
+        job_log = self._context.run_log_store.create_job_log()
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )
+        self.spin_container()
+
+    def execute_job(self, job: BaseTaskType, catalog_settings=Optional[List[str]]):
+        """
+        Focusses on execution of the job.
+        """
+        self._use_volumes()
+        logger.info("Trying to execute job")
+
+        job_log = self._context.run_log_store.get_job_log(run_id=self._context.run_id)
+
+        attempt_log = job.execute_command(
+            attempt_number=self.step_attempt_number,
+            mock=self.mock,
+        )
+
+        job_log.status = attempt_log.status
+        job_log.attempts.append(attempt_log)
+
+        data_catalogs_put: Optional[List[DataCatalog]] = self._sync_catalog(
+            catalog_settings=catalog_settings
+        )
+        logger.debug(f"data_catalogs_put: {data_catalogs_put}")
+
+        job_log.add_data_catalogs(data_catalogs_put or [])
+
+        console.print("Summary of job")
+        console.print(job_log.get_summary())
+
+        self._context.run_log_store.add_job_log(
+            run_id=self._context.run_id, job_log=job_log
+        )
+
+    def spin_container(self):
+        """
+        This method spins up the container
+        """
+        import docker  # pylint: disable=C0415
+
+        try:
+            client = docker.from_env()
+            api_client = docker.APIClient()
+        except Exception as ex:
+            logger.exception("Could not get access to docker")
+            raise Exception(
+                "Could not get the docker socket file, do you have docker installed?"
+            ) from ex
+
+        try:
+            command = utils.get_job_execution_command()
+            logger.info(f"Running the command {command}")
+            print(command)
+
+            docker_image = self.docker_image
+            environment = self.environment
+
+            container = client.containers.create(
+                image=docker_image,
+                command=command,
+                auto_remove=False,
+                volumes=self._volumes,
+                network_mode="host",
+                environment=environment,
+            )
+
+            # print(container.__dict__)
+
+            container.start()
+            stream = api_client.logs(
+                container=container.id, timestamps=True, stream=True, follow=True
+            )
+            while True:
+                try:
+                    output = next(stream).decode("utf-8")
+                    output = output.strip("\r\n")
+                    logger.info(output)
+                    print(output)
+                except StopIteration:
+                    logger.info("Docker Run completed")
+                    break
+
+            exit_status = api_client.inspect_container(container.id)["State"][
+                "ExitCode"
+            ]
+
+            if self.auto_remove_container:
+                container.remove(force=True)
+
+            if exit_status != 0:
+                msg = f"Docker command failed with exit code {exit_status}"
+                raise Exception(msg)
+
+        except Exception as _e:
+            logger.exception("Problems with spinning/running the container")
+            raise _e
+
+    def _mount_volumes(self):
+        """
+        Mount the volumes for the container
+        """
+        match self._context.run_log_store.service_name:
+            case "file-system":
+                write_to = self._context.run_log_store.log_folder
+                self._volumes[str(Path(write_to).resolve())] = {
+                    "bind": f"{self._container_log_location}",
+                    "mode": "rw",
+                }
+            case "chunked-fs":
+                write_to = self._context.run_log_store.log_folder
+                self._volumes[str(Path(write_to).resolve())] = {
+                    "bind": f"{self._container_log_location}",
+                    "mode": "rw",
+                }
+
+        match self._context.catalog_handler.service_name:
+            case "file-system":
+                catalog_location = self._context.catalog_handler.catalog_location
+                self._volumes[str(Path(catalog_location).resolve())] = {
+                    "bind": f"{self._container_catalog_location}",
+                    "mode": "rw",
+                }
+
+        match self._context.secrets_handler.service_name:
+            case "dotenv":
+                secrets_location = self._context.secrets_handler.location
+                self._volumes[str(Path(secrets_location).resolve())] = {
+                    "bind": f"{self._container_secrets_location}",
+                    "mode": "ro",
+                }
+
+    def _use_volumes(self):
+        match self._context.run_log_store.service_name:
+            case "file-system":
+                self._context.run_log_store.log_folder = self._container_log_location
+            case "chunked-fs":
+                self._context.run_log_store.log_folder = self._container_log_location
+
+        match self._context.catalog_handler.service_name:
+            case "file-system":
+                self._context.catalog_handler.catalog_location = (
+                    self._container_catalog_location
+                )
+
+        match self._context.secrets_handler.service_name:
+            case "dotenv":
+                self._context.secrets_handler.location = (
+                    self._container_secrets_location
+                )

extensions/job_executor/pyproject.toml

@@ -0,0 +1,16 @@
+[project]
+name = "job_executor"
+version = "0.0.0"
+description = "Add your description here"
+readme = "README.md"
+requires-python = ">=3.10"
+dependencies = []
+
+
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+
+[tool.hatch.build.targets.wheel]
+packages = ["."]