Legacy Database

orion.storage.legacy – Legacy storage

class orion.storage.legacy.Legacy(database=None, setup=True)

Legacy protocol, store all experiments and trials inside the Database()

Parameters:

config: Dict

configuration definition passed from experiment_builder to storage factory to legacy constructor. See Database for more details

setup: bool

Setup the database (create indexes)

Attributes:

instance

Methods

count_broken_trials(experiment)	See count_broken_trials()
count_completed_trials(experiment)	See count_completed_trials()
create_experiment(config)	See create_experiment()
delete_experiment([experiment, uid])	See delete_experiment()
delete_trials([experiment, uid, where])	See delete_trials()
fetch_experiments(query[, selection])	See fetch_experiments()
fetch_lost_trials(experiment)	See fetch_lost_trials()
fetch_noncompleted_trials(experiment)	See fetch_noncompleted_trials()
fetch_pending_trials(experiment)	See fetch_pending_trials()
fetch_trials([experiment, uid])	See fetch_trials()
fetch_trials_by_status(experiment, status)	See fetch_trials_by_status()
get_trial([trial, uid])	See get_trial()
push_trial_results(trial)	See push_trial_results()
register_lie(trial)	See register_lie()
register_trial(trial)	See register_trial()
reserve_trial(experiment)	See reserve_trial()
retrieve_result(trial[, results_file])	Parse the results file that was generated by the trial process.
set_trial_status(trial, status[, heartbeat])	See set_trial_status()
update_experiment([experiment, uid, where])	See update_experiment()
update_heartbeat(trial)	Update trial’s heartbeat
update_trial([trial, uid, where])	See update_trial()
update_trials([experiment, uid, where])	See update_trials()

count_broken_trials(experiment)

See count_broken_trials()

count_completed_trials(experiment)

See count_completed_trials()

create_experiment(config)

See create_experiment()

delete_experiment(experiment=None, uid=None)

See delete_experiment()

delete_trials(experiment=None, uid=None, where=None)

See delete_trials()

fetch_experiments(query, selection=None)

See fetch_experiments()

fetch_lost_trials(experiment)

See fetch_lost_trials()

fetch_noncompleted_trials(experiment)

See fetch_noncompleted_trials()

fetch_pending_trials(experiment)

See fetch_pending_trials()

fetch_trials(experiment=None, uid=None)

See fetch_trials()

fetch_trials_by_status(experiment, status)

See fetch_trials_by_status()

get_trial(trial=None, uid=None)

See get_trial()

push_trial_results(trial)

See push_trial_results()

register_lie(trial)

See register_lie()

register_trial(trial)

See register_trial()

reserve_trial(experiment)

See reserve_trial()

retrieve_result(trial, results_file=None, **kwargs)

Parse the results file that was generated by the trial process.

Parameters:

trial: Trial

The trial object to be updated

results_file: str

the file handle to read the result from

Returns:

returns the updated trial object

Notes

This does not update the database!

set_trial_status(trial, status, heartbeat=None)

See set_trial_status()

update_experiment(experiment=None, uid=None, where=None, **kwargs)

See update_experiment()

update_heartbeat(trial)

Update trial’s heartbeat

update_trial(trial=None, uid=None, where=None, **kwargs)

See update_trial()

update_trials(experiment=None, uid=None, where=None, **kwargs)

See update_trials()

orion.storage.legacy.get_database()

Return current database

This is a wrapper around the Database Singleton object to provide better error message when it is used without being initialized.

Raises:

RuntimeError

If the underlying database was not initialized prior to calling this function

Notes

To initialize the underlying database you must first call Database(…) with the appropriate arguments for the chosen backend

orion.storage.legacy.setup_database(config=None)

Create the Database instance from a configuration.

Parameters:

config: dict

Configuration for the database backend. If not defined, global configuration is used.