Asynchronous Base Executor Class

class covalent.executor.base.AsyncBaseExecutor(*args, **kwargs)

Async base executor class to be used for defining any executor plugin. Subclassing this class will allow you to define your own executor plugin which can be used in covalent.

This is analogous to BaseExecutor except the run() method, together with the optional setup() and teardown() methods, are coroutines.

log_stdout

The path to the file to be used for redirecting stdout.

log_stderr

The path to the file to be used for redirecting stderr.

cache_dir

The location used for cached files in the executor.

time_limit

time limit for the task

retries

Number of times to retry execution upon failure

Methods:

cancel(task_metadata, job_handle)	Executor specific task cancellation method
from_dict(object_dict)	Rehydrate a dictionary representation
get_cancel_requested()	Get if the task was requested to be canceled
get_dispatch_context(dispatch_info)	Start a context manager that will be used to access the dispatch info for the executor.
run(function, args, kwargs, task_metadata)	Abstract method to run a function in the executor in async-aware manner.
set_job_handle(handle)	Save the job handle to database
setup(task_metadata)	Executor specific setup method
teardown(task_metadata)	Executor specific teardown method
to_dict()	Return a JSON-serializable dictionary representation of self
write_streams_to_file(stream_strings, …)	Write the contents of stdout and stderr to respective files.

async cancel(task_metadata, job_handle)

Executor specific task cancellation method

Arg(s)

task_metadata: Metadata associated with the task to be cancelled job_handle: Unique ID assigned to the job by the backend

Return(s)

False by default

Return type

Literal[False]

from_dict(object_dict)

Rehydrate a dictionary representation

Parameters

object_dict (dict) – a dictionary representation returned by to_dict

Return type

BaseExecutor

Returns

self

Instance attributes will be overwritten.

async get_cancel_requested()

Get if the task was requested to be canceled

Arg(s)

None

Return(s)

Whether the task has been requested to be cancelled

Return type

Any

get_dispatch_context(dispatch_info)

Start a context manager that will be used to access the dispatch info for the executor.

Parameters

dispatch_info (DispatchInfo) – The dispatch info to be used inside current context.

Return type

AbstractContextManager[DispatchInfo]

Returns

A context manager object that handles the dispatch info.

abstract async run(function, args, kwargs, task_metadata)

Abstract method to run a function in the executor in async-aware manner.

Parameters

• function (Callable) – The function to run in the executor

• args (List) – List of positional arguments to be used by the function

• kwargs (Dict) – Dictionary of keyword arguments to be used by the function.

• task_metadata (Dict) – Dictionary of metadata for the task. Current keys are dispatch_id and node_id

Returns

The result of the function execution

Return type

output

async set_job_handle(handle)

Save the job handle to database

Arg(s)

handle: JSONable type identifying the job being executed by the backend

Return(s)

Response from the listener that handles inserting the job handle to database

Return type

Any

async setup(task_metadata)

Executor specific setup method

async teardown(task_metadata)

Executor specific teardown method

to_dict()

Return a JSON-serializable dictionary representation of self

Return type

dict

async write_streams_to_file(stream_strings, filepaths, dispatch_id, results_dir)

Write the contents of stdout and stderr to respective files.

Parameters

• stream_strings (Iterable[str]) – The stream_strings to be written to files.

• filepaths (Iterable[str]) – The filepaths to be used for writing the streams.

• dispatch_id (str) – The ID of the dispatch which initiated the request.

• results_dir (str) – The location of the results directory.

This uses aiofiles to avoid blocking the event loop.

Return type

None