Skip to content

Latest commit

 

History

History
66 lines (41 loc) · 2.88 KB

index.rst

File metadata and controls

66 lines (41 loc) · 2.88 KB

Worker Adapters

Worker Adapters are components in Scaler that handle the actual provisioning and destruction of worker resources. They act as the bridge between Scaler's scaling policies and various infrastructure providers (e.g., local processes, cloud instances, container orchestrators).

Note

For more details on how to configure Scaler, see the :doc:`../configuration` section.

Note

By default, the scheduler starts with the no scaling policy, meaning no workers are provisioned automatically. A worker adapter will connect to the scheduler but will not receive any scaling commands until a policy that performs scaling is configured. To enable auto-scaling, pass a --policy-content (-pc) flag to the scheduler.

Enabling Auto-Scaling

To enable auto-scaling with a worker adapter, configure the scheduler's scaling policy using the --policy-content (-pc) flag. For example, to use the vanilla scaler:

scaler_scheduler tcp://127.0.0.1:8516 -pc "allocate=even_load; scaling=vanilla"

Once the scheduler is running with this policy, start a worker adapter (e.g., the Native adapter):

scaler_worker_adapter_native tcp://127.0.0.1:8516 --max-workers 8

The vanilla policy will then automatically scale workers up and down based on the task-to-worker ratio. For a full description of available scaling policies and their parameters, see :doc:`../scaling`.

Adapters Overview

Scaler provides several worker adapters to support different execution environments.

Native

The :doc:`Native <native>` worker adapter allows Scaler to dynamically provision workers as local subprocesses on the same machine. It is the simplest way to scale workloads across multiple CPU cores locally and supports dynamic auto-scaling.

Fixed Native

The :doc:`Fixed Native <fixed_native>` worker adapter spawns a static number of worker subprocesses at startup and does not support dynamic scaling. It is the underlying component used by the high-level Cluster and SchedulerClusterCombo classes.

AWS HPC

The :doc:`AWS HPC <aws_hpc/index>` worker adapter allows Scaler to offload task execution to cloud environments, currently supporting AWS Batch. It is ideal for bursting workloads to the cloud or utilizing specific hardware not available locally.

ORB (AWS EC2)

The :doc:`ORB <orb>` worker adapter allows Scaler to dynamically provision workers on AWS EC2 instances. This is ideal for scaling workloads that require significant cloud compute resources or specialized hardware like GPUs.

Common Parameters

All worker adapters share a set of :doc:`common configuration parameters <common_parameters>` for networking, worker behavior, and logging.

.. toctree::
    :hidden:

    native
    fixed_native
    orb
    aws_hpc/index
    common_parameters