adaptivemd.Scheduler¶

class adaptivemd.Scheduler(resource, queue=None, runtime=240, cores=1)[source]¶

Class to handle task execution on a resource

Notes

In RP this would correspond to a Pilot with a UnitManager

Variables:	project (Project) – a back reference to the project that uses this scheduler tasks (dict uid : Task) – dict that references all running task by the associated CU.uid wrapper (Task) – a wrapping task that contains additional commands to be executed around each task running on that scheduler. It usually contains adding certain paths, etc.
Parameters:	resource (Resource) – a Resource where this scheduler works on queue (str) – the name of the queue to be used for pilot creation runtime (int) – max runtime in minutes for the created pilot cores – number of used cores to be used in the created pilot

__init__(resource, queue=None, runtime=240, cores=1)[source]¶

Parameters:	resource (Resource) – a Resource where this scheduler works on queue (str) – the name of the queue to be used for pilot creation runtime (int) – max runtime in minutes for the created pilot cores – number of used cores to be used in the created pilot

Methods

__init__(resource[, queue, runtime, cores])

param resource:	a Resource where this scheduler works on

add_event(event)

cancel_events() Remove all pending events and stop them from further task execution

change_state(new_state)

enter([project]) Call a preparations to use a scheduler

exit() Shut down the scheduler

flatten_location(obj)

get_path(f) Get the schedulers representation of the path in Location object

on(condition) Shortcut for creation and appending of a new Event

remove_task(task)

replace_prefix(path) Interprete adaptive paths and replace prefixes with real os paths

shut_down([wait_to_finish]) Do a controlled shutdown.

stage_generators() Prepare files and folder for all generators

stage_in(staging)

submit(submission) Submit a task in form of an event, a task or an task-like object

trigger() Trigger a check of state changes that leads to task execution

unroll_staging_path(location) Convert a staging location into an adaptiveMD location

wait() Wait until no more units are running and hence no more state changes

Attributes

`folder_name`
`generators`	Return the generators of the attached project
`is_idle`
`staging_area_location`	Return the path to the staging area used by this scheduler

__init__(resource, queue=None, runtime=240, cores=1)[source]

Parameters:	resource (Resource) – a Resource where this scheduler works on queue (str) – the name of the queue to be used for pilot creation runtime (int) – max runtime in minutes for the created pilot cores – number of used cores to be used in the created pilot

staging_area_location¶: Return the path to the staging area used by this scheduler

generators¶

Return the generators of the attached project

Returns:
Return type:	list of TaskGenerator

get_path(f)[source]¶

Get the schedulers representation of the path in Location object

Parameters:	f (Location) – the location object
Returns:	a real file path
Return type:	str

unroll_staging_path(location)[source]¶

Convert a staging location into an adaptiveMD location

Parameters:	location (Location) – the location to the changed

enter(project=None)[source]¶

Call a preparations to use a scheduler

Parameters:	project (Project) – the project the worker should execute for

exit()[source]¶: Shut down the scheduler

stage_generators()[source]¶: Prepare files and folder for all generators

submit(submission)[source]¶

Submit a task in form of an event, a task or an task-like object

Parameters:	submission ((list of) [Task or object or Event]) –
Returns:	the list of tasks actually executed after looking at all objects
Return type:	list of Task

trigger()[source]¶: Trigger a check of state changes that leads to task execution

shut_down(wait_to_finish=True)[source]¶

Do a controlled shutdown. Cancel all units and wait until they finish.

Parameters:	wait_to_finish (bool) – if True default the function will block until all tasks report finish

on(condition)[source]¶

Shortcut for creation and appending of a new Event

Parameters:	condition (Condition) –
Returns:
Return type:	Event

wait()[source]¶: Wait until no more units are running and hence no more state changes

cancel_events()[source]¶: Remove all pending events and stop them from further task execution

replace_prefix(path)[source]¶

Interprete adaptive paths and replace prefixes with real os paths

Parameters:	path (str) – the path with an adaptiveMD prefix
Returns:	the path without any adaptiveMD prefixes
Return type:	str