perun.monitoring.subprocess

Perun subprocess module.

Attributes

log

Functions

prepSensors(...)	Prepare sensors for monitoring.
createNode(→ perun.data_model.data.DataNode)	Create a data node from the sensor data.
perunSubprocess(→ None)	Parallel function that samples energy values from hardware libraries.

Module Contents

perun.monitoring.subprocess.log

perun.monitoring.subprocess.prepSensors(backends: dict[str, perun.backend.Backend], l_assigned_sensors: dict[str, tuple]) → tuple[perun.data_model.measurement_type.MetricMetaData, list[perun.data_model.sensor.Sensor]]

Prepare sensors for monitoring.

Parameters:

• backends (dict[str, Backend]) – A dictionary of backends.

• l_assigned_sensors (dict[str, tuple]) – A dictionary of sensor configurations.

Returns:

A tuple containing the following: - timesteps (list[int]): A list of timesteps. - t_metadata (MetricMetaData): Metadata for the metrics. - rawValues (list[list[Number]]): A list of raw sensor values. - lSensors (list[Sensor]): A list of sensors.

Return type:

tuple[list[int], MetricMetaData, list[list[Number]], list[Sensor]]

perun.monitoring.subprocess.createNode(timesteps: list[int], t_metadata: perun.data_model.measurement_type.MetricMetaData, rawValues: list[list[perun.data_model.measurement_type.Number]], lSensors: list[perun.data_model.sensor.Sensor], perunConfig: configparser.ConfigParser) → perun.data_model.data.DataNode

Create a data node from the sensor data.

Parameters:

• timesteps (list[int]) – List of timesteps.

• t_metadata (MetricMetaData) – Metadata for the metrics.

• rawValues (list[list[Number]]) – List of raw sensor values.

• lSensors (list[Sensor]) – A list of sensors.

• perunConfig (ConfigParser) – The perun configuration.

Returns:

A data node.

Return type:

DataNode

perun.monitoring.subprocess.perunSubprocess(queue: multiprocessing.Queue, rank: int, l_assigned_sensors: dict[str, tuple], perunConfig: configparser.ConfigParser, sp_ready_event: multiprocessing.synchronize.Event, start_event: multiprocessing.synchronize.Event, stop_event: multiprocessing.synchronize.Event, close_event: multiprocessing.synchronize.Event, sampling_period: float, live_callback_inits: dict[str, Callable[[], Callable[[dict[str, perun.data_model.measurement_type.Number]], None]]]) → None

Parallel function that samples energy values from hardware libraries.

Parameters:

• queue (Queue) – Multiprocessing Queue object where the results are sent after finish

• rank (int) – Local MPI Rank

• l_assigned_sensors (Dict[str, Tuple]) – Local MPI rank sensor configuration

• perunConfig (ConfigParser) – Global perun configuration

• sp_ready_event (EventClass) – Indicates monitoring supbrocess is ready, multiprocessing module

• start_event (EventClass) – Indicates start of the monitored application

• stop_event (EventClass) – Indicates the stop of a monitored application

• close_event (EventClass) – Indicates that perun is closing, and the subprocess needs to close

• sampling_period (float) – Sampling period in seconds

• live_callback_inits (dict[str, Callable[[], Callable[[str, Number], None]]]) – Dictionary of live callback initializers, where the key is the name of the callback and the value is a function that returns a callable that accepts metric identifier and value.