glance.async_ package

Subpackages

Submodules

Module contents

class glance.async_.EventletThreadPoolModel(size=None)[source]

Bases: glance.async_.ThreadPoolModel

A ThreadPoolModel suitable for use with evenlet/greenthreads.

DEFAULTSIZE = 1024
static get_threadpool_executor_class()[source]

Returns a futurist.ThreadPoolExecutor class.

class glance.async_.NativeThreadPoolModel(size=None)[source]

Bases: glance.async_.ThreadPoolModel

A ThreadPoolModel suitable for use with native threads.

DEFAULTSIZE = 16
static get_threadpool_executor_class()[source]

Returns a futurist.ThreadPoolExecutor class.

class glance.async_.TaskExecutor(context, task_repo, image_repo, image_factory, admin_repo=None)[source]

Bases: object

Base class for Asynchronous task executors. It does not support the execution mechanism.

Provisions the extensible classes with necessary variables to utilize important Glance modules like, context, task_repo, image_repo, image_factory.

Note:

It also gives abstraction for the standard pre-processing and post-processing operations to be executed by a task. These may include validation checks, security checks, introspection, error handling etc. The aim is to give developers an abstract sense of the execution pipeline logic.

Args:
context: glance.context.RequestContext object for AuthZ and AuthN

checks

task_repo: glance.db.TaskRepo object which acts as a translator for

glance.domain.Task and glance.domain.TaskStub objects into ORM semantics

image_repo: glance.db.ImageRepo object which acts as a translator for

glance.domain.Image object into ORM semantics

image_factory: glance.domain.ImageFactory object to be used for

creating new images for certain types of tasks viz. import, cloning

admin_repo: glance.db.ImageRepo object which acts as a translator for

glance.domain.Image object into ORM semantics, but with an admin context (optional)

begin_processing(task_id)[source]
class glance.async_.ThreadPoolModel(size=None)[source]

Bases: object

Base class for an abstract ThreadPool.

Do not instantiate this directly, use one of the concrete implementations.

DEFAULTSIZE = 1
static get_threadpool_executor_class()[source]

Returns a futurist.ThreadPoolExecutor class.

map(fn, iterable)[source]

Map a function to each value in an iterable.

This spawns a thread for each item in the provided iterable, generating results in the same order. Each item will spawn a thread, and each may run in parallel up to the limit of the pool.

Parameters
  • fn – A function to work on each item

  • iterable – A sequence of items to process

Returns

A generator of results in the same order

spawn(fn, *args, **kwargs)[source]

Spawn a function with args using the thread pool.

glance.async_.get_threadpool_model()[source]

Returns the system-wide threadpool model class.

This must be called after set_threadpool_model() whenever some code needs to know what the threadpool implementation is.

This may only be called after set_threadpool_model() has been called to set the desired threading mode. If it is called before the model is set, it will raise AssertionError. This would likely be the case if this got run in a test before the model was initialized, or if glance modules that use threading were imported and run from some other code without setting the model first.

Raises

AssertionError if the model has not yet been set.

glance.async_.set_threadpool_model(thread_type)[source]

Set the system-wide threadpool model.

This sets the type of ThreadPoolModel to use globally in the process. It should be called very early in init, and only once.

Parameters

thread_type – A string indicating the threading type in use, either “eventlet” or “native”

Raises

RuntimeError if the model is already set or some thread_type other than one of the supported ones is provided.