Hyperai
zenml.integrations.hyperai
special
Initialization of the HyperAI integration.
HyperAIIntegration (Integration)
Definition of HyperAI integration for ZenML.
Source code in zenml/integrations/hyperai/__init__.py
class HyperAIIntegration(Integration):
"""Definition of HyperAI integration for ZenML."""
NAME = HYPERAI
REQUIREMENTS = [
"paramiko>=3.4.0",
]
@classmethod
def activate(cls) -> None:
"""Activates the integration."""
from zenml.integrations.hyperai import service_connectors # noqa
@classmethod
def flavors(cls) -> List[Type[Flavor]]:
"""Declare the stack component flavors for the HyperAI integration.
Returns:
List of stack component flavors for this integration.
"""
from zenml.integrations.hyperai.flavors import (
HyperAIOrchestratorFlavor
)
return [HyperAIOrchestratorFlavor]
activate()
classmethod
Activates the integration.
Source code in zenml/integrations/hyperai/__init__.py
@classmethod
def activate(cls) -> None:
"""Activates the integration."""
from zenml.integrations.hyperai import service_connectors # noqa
flavors()
classmethod
Declare the stack component flavors for the HyperAI integration.
Returns:
| Type | Description |
|---|---|
List[Type[zenml.stack.flavor.Flavor]] |
List of stack component flavors for this integration. |
Source code in zenml/integrations/hyperai/__init__.py
@classmethod
def flavors(cls) -> List[Type[Flavor]]:
"""Declare the stack component flavors for the HyperAI integration.
Returns:
List of stack component flavors for this integration.
"""
from zenml.integrations.hyperai.flavors import (
HyperAIOrchestratorFlavor
)
return [HyperAIOrchestratorFlavor]
flavors
special
HyperAI integration flavors.
hyperai_orchestrator_flavor
Implementation of the ZenML HyperAI orchestrator.
HyperAIOrchestratorConfig (BaseOrchestratorConfig, HyperAIOrchestratorSettings)
pydantic-model
Configuration for the HyperAI orchestrator.
Attributes:
| Name | Type | Description |
|---|---|---|
container_registry_autologin |
bool |
If True, the orchestrator will attempt to
automatically log in to the container registry specified in the stack
configuration on the HyperAI instance. This is useful if the container
registry requires authentication and the HyperAI instance has not been
manually logged in to the container registry. Defaults to |
automatic_cleanup_pipeline_files |
bool |
If True, the orchestrator will
automatically clean up old pipeline files that are on the HyperAI
instance. Pipeline files will be cleaned up if they are 7 days old or
older. Defaults to |
gpu_enabled_in_container |
bool |
If True, the orchestrator will enable GPU
support in the Docker container that runs on the HyperAI instance.
Defaults to |
Source code in zenml/integrations/hyperai/flavors/hyperai_orchestrator_flavor.py
class HyperAIOrchestratorConfig( # type: ignore[misc] # https://github.com/pydantic/pydantic/issues/4173
BaseOrchestratorConfig, HyperAIOrchestratorSettings
):
"""Configuration for the HyperAI orchestrator.
Attributes:
container_registry_autologin: If True, the orchestrator will attempt to
automatically log in to the container registry specified in the stack
configuration on the HyperAI instance. This is useful if the container
registry requires authentication and the HyperAI instance has not been
manually logged in to the container registry. Defaults to `False`.
automatic_cleanup_pipeline_files: If True, the orchestrator will
automatically clean up old pipeline files that are on the HyperAI
instance. Pipeline files will be cleaned up if they are 7 days old or
older. Defaults to `True`.
gpu_enabled_in_container: If True, the orchestrator will enable GPU
support in the Docker container that runs on the HyperAI instance.
Defaults to `True`.
"""
container_registry_autologin: bool = False
automatic_cleanup_pipeline_files: bool = True
gpu_enabled_in_container: bool = True
@property
def is_remote(self) -> bool:
"""Checks if this stack component is running remotely.
This designation is used to determine if the stack component can be
used with a local ZenML database or if it requires a remote ZenML
server.
Returns:
True if this config is for a remote component, False otherwise.
"""
return True
is_remote: bool
property
readonly
Checks if this stack component is running remotely.
This designation is used to determine if the stack component can be used with a local ZenML database or if it requires a remote ZenML server.
Returns:
| Type | Description |
|---|---|
bool |
True if this config is for a remote component, False otherwise. |
HyperAIOrchestratorFlavor (BaseOrchestratorFlavor)
Flavor for the HyperAI orchestrator.
Source code in zenml/integrations/hyperai/flavors/hyperai_orchestrator_flavor.py
class HyperAIOrchestratorFlavor(BaseOrchestratorFlavor):
"""Flavor for the HyperAI orchestrator."""
@property
def name(self) -> str:
"""Name of the orchestrator flavor.
Returns:
Name of the orchestrator flavor.
"""
return "hyperai"
@property
def service_connector_requirements(
self,
) -> Optional[ServiceConnectorRequirements]:
"""Service connector resource requirements for service connectors.
Specifies resource requirements that are used to filter the available
service connector types that are compatible with this flavor.
Returns:
Requirements for compatible service connectors, if a service
connector is required for this flavor.
"""
return ServiceConnectorRequirements(
resource_type=HYPERAI_RESOURCE_TYPE
)
@property
def docs_url(self) -> Optional[str]:
"""A url to point at docs explaining this flavor.
Returns:
A flavor docs url.
"""
return self.generate_default_docs_url()
@property
def sdk_docs_url(self) -> Optional[str]:
"""A url to point at SDK docs explaining this flavor.
Returns:
A flavor SDK docs url.
"""
return self.generate_default_sdk_docs_url()
@property
def logo_url(self) -> str:
"""A url to represent the flavor in the dashboard.
Returns:
The flavor logo.
"""
return "https://public-flavor-logos.s3.eu-central-1.amazonaws.com/connectors/hyperai/hyperai.png"
@property
def config_class(self) -> Type[BaseOrchestratorConfig]:
"""Config class for the base orchestrator flavor.
Returns:
The config class.
"""
return HyperAIOrchestratorConfig
@property
def implementation_class(self) -> Type["HyperAIOrchestrator"]:
"""Implementation class for this flavor.
Returns:
Implementation class for this flavor.
"""
from zenml.integrations.hyperai.orchestrators import (
HyperAIOrchestrator,
)
return HyperAIOrchestrator
config_class: Type[zenml.orchestrators.base_orchestrator.BaseOrchestratorConfig]
property
readonly
Config class for the base orchestrator flavor.
Returns:
| Type | Description |
|---|---|
Type[zenml.orchestrators.base_orchestrator.BaseOrchestratorConfig] |
The config class. |
docs_url: Optional[str]
property
readonly
A url to point at docs explaining this flavor.
Returns:
| Type | Description |
|---|---|
Optional[str] |
A flavor docs url. |
implementation_class: Type[HyperAIOrchestrator]
property
readonly
Implementation class for this flavor.
Returns:
| Type | Description |
|---|---|
Type[HyperAIOrchestrator] |
Implementation class for this flavor. |
logo_url: str
property
readonly
A url to represent the flavor in the dashboard.
Returns:
| Type | Description |
|---|---|
str |
The flavor logo. |
name: str
property
readonly
Name of the orchestrator flavor.
Returns:
| Type | Description |
|---|---|
str |
Name of the orchestrator flavor. |
sdk_docs_url: Optional[str]
property
readonly
A url to point at SDK docs explaining this flavor.
Returns:
| Type | Description |
|---|---|
Optional[str] |
A flavor SDK docs url. |
service_connector_requirements: Optional[zenml.models.v2.misc.service_connector_type.ServiceConnectorRequirements]
property
readonly
Service connector resource requirements for service connectors.
Specifies resource requirements that are used to filter the available service connector types that are compatible with this flavor.
Returns:
| Type | Description |
|---|---|
Optional[zenml.models.v2.misc.service_connector_type.ServiceConnectorRequirements] |
Requirements for compatible service connectors, if a service connector is required for this flavor. |
HyperAIOrchestratorSettings (BaseSettings)
pydantic-model
HyperAI orchestrator settings.
Attributes:
| Name | Type | Description |
|---|---|---|
mounts_from_to |
Dict[str, str] |
A dictionary mapping from paths on the HyperAI instance to paths within the Docker container. This allows users to mount directories from the HyperAI instance into the Docker container that runs on it. |
Source code in zenml/integrations/hyperai/flavors/hyperai_orchestrator_flavor.py
class HyperAIOrchestratorSettings(BaseSettings):
"""HyperAI orchestrator settings.
Attributes:
mounts_from_to: A dictionary mapping from paths on the HyperAI instance
to paths within the Docker container. This allows users to mount
directories from the HyperAI instance into the Docker container that runs
on it.
"""
mounts_from_to: Dict[str, str] = {}
orchestrators
special
HyperAI orchestrator.
hyperai_orchestrator
Implementation of the ZenML HyperAI orchestrator.
HyperAIOrchestrator (ContainerizedOrchestrator)
Orchestrator responsible for running pipelines on HyperAI instances.
Source code in zenml/integrations/hyperai/orchestrators/hyperai_orchestrator.py
class HyperAIOrchestrator(ContainerizedOrchestrator):
"""Orchestrator responsible for running pipelines on HyperAI instances."""
@property
def config(self) -> HyperAIOrchestratorConfig:
"""Returns the `HyperAIOrchestratorConfig` config.
Returns:
The configuration.
"""
return cast(HyperAIOrchestratorConfig, self._config)
@property
def settings_class(self) -> Optional[Type["BaseSettings"]]:
"""Settings class for the HyperAI orchestrator.
Returns:
The settings class.
"""
return HyperAIOrchestratorSettings
@property
def validator(self) -> Optional[StackValidator]:
"""Ensures there is an image builder in the stack.
Returns:
A `StackValidator` instance.
"""
return StackValidator(
required_components={
StackComponentType.CONTAINER_REGISTRY,
StackComponentType.IMAGE_BUILDER,
}
)
def get_orchestrator_run_id(self) -> str:
"""Returns the active orchestrator run id.
Raises:
RuntimeError: If the environment variable specifying the run id
is not set.
Returns:
The orchestrator run id.
"""
try:
return os.environ[ENV_ZENML_HYPERAI_RUN_ID]
except KeyError:
raise RuntimeError(
"Unable to read run id from environment variable "
f"{ENV_ZENML_HYPERAI_RUN_ID}."
)
def _validate_mount_path(self, path: str) -> str:
"""Validates if a given string is in a valid path format.
Args:
path: The path to be validated.
Returns:
The path in a valid format.
Raises:
RuntimeError: If the path is not in a valid format.
"""
# Define a regular expression pattern to match a valid path format
pattern = r'^(?:[a-zA-Z]:\\(\\[^\\/:*?"<>|]*)*$|^/([^\0]*)*$)'
if bool(re.match(pattern, path)):
return path
else:
raise RuntimeError(
f"Path '{path}' is not in a valid format, so a mount cannot be established."
)
def _escape_shell_command(self, command: str) -> str:
"""Escapes a shell command.
Args:
command: The command to escape.
Returns:
The escaped command.
"""
return quote(command)
def prepare_or_run_pipeline(
self,
deployment: "PipelineDeploymentResponse",
stack: "Stack",
environment: Dict[str, str],
) -> Any:
"""Sequentially runs all pipeline steps in Docker containers.
Assumes that:
- A HyperAI (hyperai.ai) instance is running on the configured IP address.
- The HyperAI instance has been configured to allow SSH connections from the
machine running the pipeline.
- Docker and Docker Compose are installed on the HyperAI instance.
- A key pair has been generated and the public key has been added to the
HyperAI instance's `authorized_keys` file.
- The private key is available in a HyperAI service connector linked to this
orchestrator.
Args:
deployment: The pipeline deployment to prepare or run.
stack: The stack the pipeline will run on.
environment: Environment variables to set in the orchestration
environment.
Raises:
RuntimeError: If a step fails.
"""
from zenml.integrations.hyperai.service_connectors.hyperai_service_connector import (
HyperAIServiceConnector,
)
# Basic Docker Compose definition
compose_definition: Dict[str, Any] = {"version": "3", "services": {}}
# Get deployment id
deployment_id = deployment.id
# Set environment
os.environ[ENV_ZENML_HYPERAI_RUN_ID] = str(deployment_id)
environment[ENV_ZENML_HYPERAI_RUN_ID] = str(deployment_id)
# Add each step as a service to the Docker Compose definition
logger.info("Preparing pipeline steps for deployment.")
for step_name, step in deployment.step_configurations.items():
# Get image
image = self.get_image(deployment=deployment, step_name=step_name)
# Get settings
step_settings = cast(
HyperAIOrchestratorSettings, self.get_settings(step)
)
# Define container name as combination between deployment id and step name
container_name = f"{deployment_id}-{step_name}"
# Make Compose service definition for step
compose_definition["services"][container_name] = {
"image": image,
"container_name": container_name,
"network_mode": "host",
"entrypoint": StepEntrypointConfiguration.get_entrypoint_command(),
"command": StepEntrypointConfiguration.get_entrypoint_arguments(
step_name=step_name, deployment_id=deployment.id
),
"volumes": [
"{}:{}".format(
self._validate_mount_path(mount_from),
self._validate_mount_path(mount_to),
)
for mount_from, mount_to in step_settings.mounts_from_to.items()
],
}
# Depending on GPU setting, add GPU support to service definition
if self.config.gpu_enabled_in_container:
compose_definition["services"][container_name]["deploy"] = {
"resources": {
"reservations": {
"devices": [
{"driver": "nvidia", "capabilities": ["gpu"]}
]
}
}
}
# Depending on whether it is a scheduled or a realtime pipeline, add
# potential .env file to service definition for deployment ID override.
if deployment.schedule:
# drop ZENML_HYPERAI_ORCHESTRATOR_RUN_ID from environment
del environment[ENV_ZENML_HYPERAI_RUN_ID]
compose_definition["services"][container_name]["env_file"] = [
".env"
]
compose_definition["services"][container_name][
"environment"
] = environment
# Add dependency on upstream steps if applicable
upstream_steps = step.spec.upstream_steps
for upstream_step_name in upstream_steps:
upstream_container_name = (
f"{deployment_id}-{upstream_step_name}"
)
compose_definition["services"][container_name][
"depends_on"
] = {
upstream_container_name: {
"condition": "service_completed_successfully"
}
}
# Convert into yaml
logger.info("Finalizing Docker Compose definition.")
compose_definition_yaml: str = yaml.dump(compose_definition)
# Connect to configured HyperAI instance
logger.info(
"Connecting to HyperAI instance and placing Docker Compose file."
)
paramiko_client: paramiko.SSHClient
if connector := self.get_connector():
paramiko_client = connector.connect()
if paramiko_client is None:
raise RuntimeError(
"Expected to receive a `paramiko.SSHClient` object from the "
"linked connector, but got `None`. This likely originates from "
"a misconfigured service connector, typically caused by a wrong "
"SSH key type being selected. Please check your "
"`hyperai_orchestrator` configuration and make sure that the "
"`ssh_key_type` of its connected service connector is set to the "
"correct value."
)
elif not isinstance(paramiko_client, paramiko.SSHClient):
raise RuntimeError(
f"Expected to receive a `paramiko.SSHClient` object from the "
f"linked connector, but got type `{type(paramiko_client)}`."
)
else:
raise RuntimeError(
"You must link a HyperAI service connector to the orchestrator."
)
# Get container registry autologin setting
if self.config.container_registry_autologin:
logger.info(
"Attempting to automatically log in to container registry used by stack."
)
# Select stack container registry
container_registry = stack.container_registry
# Raise error if no container registry is found
if not container_registry:
raise RuntimeError(
"Unable to find container registry in stack."
)
# Get container registry credentials from its config
credentials = container_registry.credentials
if credentials is None:
raise RuntimeError(
"The container registry in the active stack has no "
"credentials or service connector configured, but the "
"HyperAI orchestrator is set to autologin to the container "
"registry. Please configure the container registry with "
"credentials or turn off the `container_registry_autologin` "
"setting in the HyperAI orchestrator configuration."
)
container_registry_url = container_registry.config.uri
(
container_registry_username,
container_registry_password,
) = credentials
# Escape inputs
container_registry_username = self._escape_shell_command(
container_registry_username
)
container_registry_url = self._escape_shell_command(
container_registry_url
)
# Log in to container registry using --password-stdin
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"docker login -u {container_registry_username} "
f"--password-stdin {container_registry_url}"
)
# Send the password to stdin
stdin.channel.send(
f"{container_registry_password}\n".encode("utf-8")
)
stdin.channel.shutdown_write()
# Log stdout
for line in stdout.readlines():
logger.info(line)
# Get username from connector
assert isinstance(connector, HyperAIServiceConnector)
username = connector.config.username
# Set up pipeline-runs directory if it doesn't exist
nonscheduled_directory_name = self._escape_shell_command(
f"/home/{username}/pipeline-runs"
)
directory_name = (
nonscheduled_directory_name
if not deployment.schedule
else self._escape_shell_command(
f"/home/{username}/scheduled-pipeline-runs"
)
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"mkdir -p {directory_name}"
)
# Get pipeline run id and create directory for it
orchestrator_run_id = self.get_orchestrator_run_id()
directory_name = self._escape_shell_command(
f"{directory_name}/{orchestrator_run_id}"
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"mkdir -p {directory_name}"
)
# Remove all folders from nonscheduled pipelines if they are 7 days old or older
if self.config.automatic_cleanup_pipeline_files:
logger.info(
"Cleaning up old pipeline files on HyperAI instance. This may take a while."
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"find {nonscheduled_directory_name} -type d -ctime +7 -exec rm -rf {{}} +"
)
# Create temporary file and write Docker Compose file to it
with tempfile.NamedTemporaryFile(mode="w", delete=True) as f:
# Write Docker Compose file to temporary file
with f.file as f_:
f_.write(compose_definition_yaml)
# Scp Docker Compose file to HyperAI instance
try:
scp_client = paramiko_client.open_sftp()
scp_client.put(f.name, f"{directory_name}/docker-compose.yaml")
scp_client.close()
except FileNotFoundError:
raise RuntimeError(
"Failed to write Docker Compose file to HyperAI instance. Does the user have permissions to write?"
)
# Run or schedule Docker Compose file depending on settings
if not deployment.schedule:
logger.info(
"Starting ZenML pipeline on HyperAI instance. Depending on the size of your container image, this may take a while..."
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"cd {directory_name} && docker compose up -d"
)
# Log errors in case of failure
for line in stderr.readlines():
logger.info(line)
else:
# Get cron expression for scheduled pipeline
cron_expression = deployment.schedule.cron_expression
if not cron_expression:
raise RuntimeError(
"A cron expression is required for scheduled pipelines."
)
expected_cron_pattern = r"^(?:(?:[0-9]|[1-5][0-9]|60)(?:,(?:[0-9]|[1-5][0-9]|60))*|[*](?:\/[1-9][0-9]*)?)(?:[ \t]+(?:(?:[0-9]|[0-5][0-9]|60)(?:,(?:[0-9]|[0-5][0-9]|60))*|[*](?:\/[1-9][0-9]*)?)){4}$"
if not re.match(expected_cron_pattern, cron_expression):
raise RuntimeError(
f"The cron expression '{cron_expression}' is not in a valid format."
)
# Log about scheduling
logger.info("Scheduling ZenML pipeline on HyperAI instance.")
logger.info(f"Cron expression: {cron_expression}")
# Create cron job for scheduled pipeline on HyperAI instance
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"(crontab -l ; echo '{cron_expression} cd {directory_name} && echo {ENV_ZENML_HYPERAI_RUN_ID}=\"{deployment_id}_$(date +\%s)\" > .env && docker compose up -d') | crontab -"
)
logger.info("Pipeline scheduled successfully.")
config: HyperAIOrchestratorConfig
property
readonly
Returns the HyperAIOrchestratorConfig config.
Returns:
| Type | Description |
|---|---|
HyperAIOrchestratorConfig |
The configuration. |
settings_class: Optional[Type[BaseSettings]]
property
readonly
Settings class for the HyperAI orchestrator.
Returns:
| Type | Description |
|---|---|
Optional[Type[BaseSettings]] |
The settings class. |
validator: Optional[zenml.stack.stack_validator.StackValidator]
property
readonly
Ensures there is an image builder in the stack.
Returns:
| Type | Description |
|---|---|
Optional[zenml.stack.stack_validator.StackValidator] |
A |
get_orchestrator_run_id(self)
Returns the active orchestrator run id.
Exceptions:
| Type | Description |
|---|---|
RuntimeError |
If the environment variable specifying the run id is not set. |
Returns:
| Type | Description |
|---|---|
str |
The orchestrator run id. |
Source code in zenml/integrations/hyperai/orchestrators/hyperai_orchestrator.py
def get_orchestrator_run_id(self) -> str:
"""Returns the active orchestrator run id.
Raises:
RuntimeError: If the environment variable specifying the run id
is not set.
Returns:
The orchestrator run id.
"""
try:
return os.environ[ENV_ZENML_HYPERAI_RUN_ID]
except KeyError:
raise RuntimeError(
"Unable to read run id from environment variable "
f"{ENV_ZENML_HYPERAI_RUN_ID}."
)
prepare_or_run_pipeline(self, deployment, stack, environment)
Sequentially runs all pipeline steps in Docker containers.
Assumes that:
- A HyperAI (hyperai.ai) instance is running on the configured IP address.
- The HyperAI instance has been configured to allow SSH connections from the
machine running the pipeline.
- Docker and Docker Compose are installed on the HyperAI instance.
- A key pair has been generated and the public key has been added to the
HyperAI instance's authorized_keys file.
- The private key is available in a HyperAI service connector linked to this
orchestrator.
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
deployment |
PipelineDeploymentResponse |
The pipeline deployment to prepare or run. |
required |
stack |
Stack |
The stack the pipeline will run on. |
required |
environment |
Dict[str, str] |
Environment variables to set in the orchestration environment. |
required |
Exceptions:
| Type | Description |
|---|---|
RuntimeError |
If a step fails. |
Source code in zenml/integrations/hyperai/orchestrators/hyperai_orchestrator.py
def prepare_or_run_pipeline(
self,
deployment: "PipelineDeploymentResponse",
stack: "Stack",
environment: Dict[str, str],
) -> Any:
"""Sequentially runs all pipeline steps in Docker containers.
Assumes that:
- A HyperAI (hyperai.ai) instance is running on the configured IP address.
- The HyperAI instance has been configured to allow SSH connections from the
machine running the pipeline.
- Docker and Docker Compose are installed on the HyperAI instance.
- A key pair has been generated and the public key has been added to the
HyperAI instance's `authorized_keys` file.
- The private key is available in a HyperAI service connector linked to this
orchestrator.
Args:
deployment: The pipeline deployment to prepare or run.
stack: The stack the pipeline will run on.
environment: Environment variables to set in the orchestration
environment.
Raises:
RuntimeError: If a step fails.
"""
from zenml.integrations.hyperai.service_connectors.hyperai_service_connector import (
HyperAIServiceConnector,
)
# Basic Docker Compose definition
compose_definition: Dict[str, Any] = {"version": "3", "services": {}}
# Get deployment id
deployment_id = deployment.id
# Set environment
os.environ[ENV_ZENML_HYPERAI_RUN_ID] = str(deployment_id)
environment[ENV_ZENML_HYPERAI_RUN_ID] = str(deployment_id)
# Add each step as a service to the Docker Compose definition
logger.info("Preparing pipeline steps for deployment.")
for step_name, step in deployment.step_configurations.items():
# Get image
image = self.get_image(deployment=deployment, step_name=step_name)
# Get settings
step_settings = cast(
HyperAIOrchestratorSettings, self.get_settings(step)
)
# Define container name as combination between deployment id and step name
container_name = f"{deployment_id}-{step_name}"
# Make Compose service definition for step
compose_definition["services"][container_name] = {
"image": image,
"container_name": container_name,
"network_mode": "host",
"entrypoint": StepEntrypointConfiguration.get_entrypoint_command(),
"command": StepEntrypointConfiguration.get_entrypoint_arguments(
step_name=step_name, deployment_id=deployment.id
),
"volumes": [
"{}:{}".format(
self._validate_mount_path(mount_from),
self._validate_mount_path(mount_to),
)
for mount_from, mount_to in step_settings.mounts_from_to.items()
],
}
# Depending on GPU setting, add GPU support to service definition
if self.config.gpu_enabled_in_container:
compose_definition["services"][container_name]["deploy"] = {
"resources": {
"reservations": {
"devices": [
{"driver": "nvidia", "capabilities": ["gpu"]}
]
}
}
}
# Depending on whether it is a scheduled or a realtime pipeline, add
# potential .env file to service definition for deployment ID override.
if deployment.schedule:
# drop ZENML_HYPERAI_ORCHESTRATOR_RUN_ID from environment
del environment[ENV_ZENML_HYPERAI_RUN_ID]
compose_definition["services"][container_name]["env_file"] = [
".env"
]
compose_definition["services"][container_name][
"environment"
] = environment
# Add dependency on upstream steps if applicable
upstream_steps = step.spec.upstream_steps
for upstream_step_name in upstream_steps:
upstream_container_name = (
f"{deployment_id}-{upstream_step_name}"
)
compose_definition["services"][container_name][
"depends_on"
] = {
upstream_container_name: {
"condition": "service_completed_successfully"
}
}
# Convert into yaml
logger.info("Finalizing Docker Compose definition.")
compose_definition_yaml: str = yaml.dump(compose_definition)
# Connect to configured HyperAI instance
logger.info(
"Connecting to HyperAI instance and placing Docker Compose file."
)
paramiko_client: paramiko.SSHClient
if connector := self.get_connector():
paramiko_client = connector.connect()
if paramiko_client is None:
raise RuntimeError(
"Expected to receive a `paramiko.SSHClient` object from the "
"linked connector, but got `None`. This likely originates from "
"a misconfigured service connector, typically caused by a wrong "
"SSH key type being selected. Please check your "
"`hyperai_orchestrator` configuration and make sure that the "
"`ssh_key_type` of its connected service connector is set to the "
"correct value."
)
elif not isinstance(paramiko_client, paramiko.SSHClient):
raise RuntimeError(
f"Expected to receive a `paramiko.SSHClient` object from the "
f"linked connector, but got type `{type(paramiko_client)}`."
)
else:
raise RuntimeError(
"You must link a HyperAI service connector to the orchestrator."
)
# Get container registry autologin setting
if self.config.container_registry_autologin:
logger.info(
"Attempting to automatically log in to container registry used by stack."
)
# Select stack container registry
container_registry = stack.container_registry
# Raise error if no container registry is found
if not container_registry:
raise RuntimeError(
"Unable to find container registry in stack."
)
# Get container registry credentials from its config
credentials = container_registry.credentials
if credentials is None:
raise RuntimeError(
"The container registry in the active stack has no "
"credentials or service connector configured, but the "
"HyperAI orchestrator is set to autologin to the container "
"registry. Please configure the container registry with "
"credentials or turn off the `container_registry_autologin` "
"setting in the HyperAI orchestrator configuration."
)
container_registry_url = container_registry.config.uri
(
container_registry_username,
container_registry_password,
) = credentials
# Escape inputs
container_registry_username = self._escape_shell_command(
container_registry_username
)
container_registry_url = self._escape_shell_command(
container_registry_url
)
# Log in to container registry using --password-stdin
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"docker login -u {container_registry_username} "
f"--password-stdin {container_registry_url}"
)
# Send the password to stdin
stdin.channel.send(
f"{container_registry_password}\n".encode("utf-8")
)
stdin.channel.shutdown_write()
# Log stdout
for line in stdout.readlines():
logger.info(line)
# Get username from connector
assert isinstance(connector, HyperAIServiceConnector)
username = connector.config.username
# Set up pipeline-runs directory if it doesn't exist
nonscheduled_directory_name = self._escape_shell_command(
f"/home/{username}/pipeline-runs"
)
directory_name = (
nonscheduled_directory_name
if not deployment.schedule
else self._escape_shell_command(
f"/home/{username}/scheduled-pipeline-runs"
)
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"mkdir -p {directory_name}"
)
# Get pipeline run id and create directory for it
orchestrator_run_id = self.get_orchestrator_run_id()
directory_name = self._escape_shell_command(
f"{directory_name}/{orchestrator_run_id}"
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"mkdir -p {directory_name}"
)
# Remove all folders from nonscheduled pipelines if they are 7 days old or older
if self.config.automatic_cleanup_pipeline_files:
logger.info(
"Cleaning up old pipeline files on HyperAI instance. This may take a while."
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"find {nonscheduled_directory_name} -type d -ctime +7 -exec rm -rf {{}} +"
)
# Create temporary file and write Docker Compose file to it
with tempfile.NamedTemporaryFile(mode="w", delete=True) as f:
# Write Docker Compose file to temporary file
with f.file as f_:
f_.write(compose_definition_yaml)
# Scp Docker Compose file to HyperAI instance
try:
scp_client = paramiko_client.open_sftp()
scp_client.put(f.name, f"{directory_name}/docker-compose.yaml")
scp_client.close()
except FileNotFoundError:
raise RuntimeError(
"Failed to write Docker Compose file to HyperAI instance. Does the user have permissions to write?"
)
# Run or schedule Docker Compose file depending on settings
if not deployment.schedule:
logger.info(
"Starting ZenML pipeline on HyperAI instance. Depending on the size of your container image, this may take a while..."
)
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"cd {directory_name} && docker compose up -d"
)
# Log errors in case of failure
for line in stderr.readlines():
logger.info(line)
else:
# Get cron expression for scheduled pipeline
cron_expression = deployment.schedule.cron_expression
if not cron_expression:
raise RuntimeError(
"A cron expression is required for scheduled pipelines."
)
expected_cron_pattern = r"^(?:(?:[0-9]|[1-5][0-9]|60)(?:,(?:[0-9]|[1-5][0-9]|60))*|[*](?:\/[1-9][0-9]*)?)(?:[ \t]+(?:(?:[0-9]|[0-5][0-9]|60)(?:,(?:[0-9]|[0-5][0-9]|60))*|[*](?:\/[1-9][0-9]*)?)){4}$"
if not re.match(expected_cron_pattern, cron_expression):
raise RuntimeError(
f"The cron expression '{cron_expression}' is not in a valid format."
)
# Log about scheduling
logger.info("Scheduling ZenML pipeline on HyperAI instance.")
logger.info(f"Cron expression: {cron_expression}")
# Create cron job for scheduled pipeline on HyperAI instance
stdin, stdout, stderr = paramiko_client.exec_command( # nosec
f"(crontab -l ; echo '{cron_expression} cd {directory_name} && echo {ENV_ZENML_HYPERAI_RUN_ID}=\"{deployment_id}_$(date +\%s)\" > .env && docker compose up -d') | crontab -"
)
logger.info("Pipeline scheduled successfully.")
service_connectors
special
HyperAI Service Connector.
hyperai_service_connector
HyperAI Service Connector.
The HyperAI Service Connector allows authenticating to HyperAI (hyperai.ai) GPU equipped instances.
HyperAIAuthenticationMethods (StrEnum)
HyperAI Authentication methods.
Source code in zenml/integrations/hyperai/service_connectors/hyperai_service_connector.py
class HyperAIAuthenticationMethods(StrEnum):
"""HyperAI Authentication methods."""
RSA_KEY_OPTIONAL_PASSPHRASE = "rsa-key"
DSA_KEY_OPTIONAL_PASSPHRASE = "dsa-key"
ECDSA_KEY_OPTIONAL_PASSPHRASE = "ecdsa-key"
ED25519_KEY_OPTIONAL_PASSPHRASE = "ed25519-key"
HyperAIConfiguration (HyperAICredentials)
pydantic-model
HyperAI client configuration.
Source code in zenml/integrations/hyperai/service_connectors/hyperai_service_connector.py
class HyperAIConfiguration(HyperAICredentials):
"""HyperAI client configuration."""
hostnames: List[str] = Field(
title="Hostnames of the supported HyperAI instances.",
)
username: str = Field(
title="Username to use to connect to the HyperAI instance.",
)
HyperAICredentials (AuthenticationConfig)
pydantic-model
HyperAI client authentication credentials.
Source code in zenml/integrations/hyperai/service_connectors/hyperai_service_connector.py
class HyperAICredentials(AuthenticationConfig):
"""HyperAI client authentication credentials."""
base64_ssh_key: SecretStr = Field(
title="SSH key (base64)",
)
ssh_passphrase: Optional[SecretStr] = Field(
default=None,
title="SSH key passphrase",
)
HyperAIServiceConnector (ServiceConnector)
pydantic-model
HyperAI service connector.
Source code in zenml/integrations/hyperai/service_connectors/hyperai_service_connector.py
class HyperAIServiceConnector(ServiceConnector):
"""HyperAI service connector."""
config: HyperAIConfiguration
@classmethod
def _get_connector_type(cls) -> ServiceConnectorTypeModel:
"""Get the service connector specification.
Returns:
The service connector specification.
"""
return HYPERAI_SERVICE_CONNECTOR_TYPE_SPEC
def _paramiko_key_type_given_auth_method(self) -> Type[paramiko.PKey]:
"""Get the Paramiko key type given the authentication method.
Returns:
The Paramiko key type.
Raises:
ValueError: If the authentication method is invalid.
"""
mapping = {
HyperAIAuthenticationMethods.RSA_KEY_OPTIONAL_PASSPHRASE: paramiko.RSAKey,
HyperAIAuthenticationMethods.DSA_KEY_OPTIONAL_PASSPHRASE: paramiko.DSSKey,
HyperAIAuthenticationMethods.ECDSA_KEY_OPTIONAL_PASSPHRASE: paramiko.ECDSAKey,
HyperAIAuthenticationMethods.ED25519_KEY_OPTIONAL_PASSPHRASE: paramiko.Ed25519Key,
}
try:
return mapping[HyperAIAuthenticationMethods(self.auth_method)]
except KeyError:
raise ValueError(
f"Invalid authentication method: {self.auth_method}"
)
def _create_paramiko_client(
self, hostname: str
) -> paramiko.client.SSHClient:
"""Create a Paramiko SSH client based on the configuration.
Args:
hostname: The hostname of the HyperAI instance.
Returns:
A Paramiko SSH client.
Raises:
AuthorizationException: If the client cannot be created.
"""
if self.config.ssh_passphrase is None:
ssh_passphrase = None
else:
ssh_passphrase = self.config.ssh_passphrase.get_secret_value()
# Connect to the HyperAI instance
try:
# Convert the SSH key from base64 to string
base64_key_value = self.config.base64_ssh_key.get_secret_value()
ssh_key = base64.b64decode(base64_key_value).decode("utf-8")
paramiko_key = None
with io.StringIO(ssh_key) as f:
paramiko_key = self._paramiko_key_type_given_auth_method().from_private_key(
f, password=ssh_passphrase
)
# Trim whitespace from the IP address
hostname = hostname.strip()
paramiko_client = paramiko.client.SSHClient()
paramiko_client.set_missing_host_key_policy(
paramiko.AutoAddPolicy() # nosec
)
paramiko_client.connect(
hostname=hostname,
username=self.config.username,
pkey=paramiko_key,
timeout=30,
)
return paramiko_client
except paramiko.ssh_exception.BadHostKeyException as e:
logger.error("Bad host key: %s", e)
except paramiko.ssh_exception.AuthenticationException as e:
logger.error("Authentication failed: %s", e)
except paramiko.ssh_exception.SSHException as e:
logger.error(
"SSH error: %s. A common cause for this error is selection of the wrong key type in your service connector.",
e,
)
except Exception as e:
logger.error(
"Unknown error while connecting to HyperAI instance: %s. Please check your network connection, IP address, and authentication details.",
e,
)
raise AuthorizationException(
"Could not create SSH client for HyperAI instance."
)
def _authorize_client(self, hostname: str) -> None:
"""Verify that the client can authenticate with the HyperAI instance.
Args:
hostname: The hostname of the HyperAI instance.
"""
logger.info("Verifying connection to HyperAI instance...")
paramiko_client = self._create_paramiko_client(hostname)
paramiko_client.close()
def _connect_to_resource(
self,
**kwargs: Any,
) -> Any:
"""Connect to a HyperAI instance. Returns an authenticated SSH client.
Args:
kwargs: Additional implementation specific keyword arguments to pass
to the session or client constructor.
Returns:
An authenticated Paramiko SSH client.
"""
logger.info("Connecting to HyperAI instance...")
assert self.resource_id is not None
paramiko_client = self._create_paramiko_client(self.resource_id)
return paramiko_client
def _configure_local_client(
self,
**kwargs: Any,
) -> None:
"""There is no local client for the HyperAI connector, so it does nothing.
Args:
kwargs: Additional implementation specific keyword arguments to pass
to the session or client constructor.
Raises:
NotImplementedError: If there is no local client for the HyperAI
connector.
"""
raise NotImplementedError(
"There is no local client for the HyperAI connector."
)
@classmethod
def _auto_configure(
cls,
auth_method: Optional[str] = None,
resource_type: Optional[str] = None,
resource_id: Optional[str] = None,
**kwargs: Any,
) -> "HyperAIServiceConnector":
"""Auto-configure the connector.
Not supported by the HyperAI connector.
Args:
auth_method: The particular authentication method to use. If not
specified, the connector implementation must decide which
authentication method to use or raise an exception.
resource_type: The type of resource to configure.
resource_id: The ID of the resource to configure. The
implementation may choose to either require or ignore this
parameter if it does not support or detect an resource type that
supports multiple instances.
kwargs: Additional implementation specific keyword arguments to use.
Raises:
NotImplementedError: If the connector auto-configuration fails or
is not supported.
"""
raise NotImplementedError(
"Auto-configuration is not supported by the HyperAI connector."
)
def _verify(
self,
resource_type: Optional[str] = None,
resource_id: Optional[str] = None,
) -> List[str]:
"""Verify that a connection can be established to the HyperAI instance.
Args:
resource_type: The type of resource to verify. Must be set to the
Docker resource type.
resource_id: The HyperAI instance to verify.
Returns:
The resource ID if the connection can be established.
Raises:
ValueError: If the resource ID is not in the list of configured
hostnames.
"""
if resource_id:
if resource_id not in self.config.hostnames:
raise ValueError(
f"The supplied hostname '{resource_id}' is not in the list "
f"of configured hostnames: {self.config.hostnames}. Please "
f"check your configuration."
)
hostnames = [resource_id]
else:
hostnames = self.config.hostnames
resources = []
for hostname in hostnames:
self._authorize_client(hostname)
resources.append(hostname)
return resources