gd_node.performance_metrics package

Submodules

gd_node.performance_metrics.collector module

Copyright (C) Mov.ai - All Rights Reserved Unauthorized copying of this file, via any medium is strictly prohibited Proprietary and confidential

Performance metrics collector for GD_Node

class gd_node.performance_metrics.collector.InfluxDBMetricsCollector(node_name: str | None = None, instance_name: str | None = None, host: str | None = None, dbname: str | None = None)

Bases: MetricsCollector

Collector that saves metrics to InfluxDB

save_metrics() → None

Write metrics to InfluxDB. Time values are stored as floats in seconds.

class gd_node.performance_metrics.collector.MetricsCollector(node_name: str | None = None, instance_name: str | None = None)

Bases: ABC

Collects and stores performance metrics for GD_Node execution. Time measurements are stored in seconds throughout the system. Only collects metrics if PLATFORM_METRICS environment variable is set.

calculate_durations() → Dict[str, float]

Calculate time durations from stored timestamps. These values are stored as floats to maintain precision.

Returns:

Dictionary of duration measurements in seconds

Return type:

Dict[str, float]

static create(*args: Any, **kwargs: Any) → NullMetricsCollector | InfluxDBMetricsCollector

Factory method that returns appropriate collector based on environment

mark_running() → None

Mark node as running after initialization is complete

mark_terminated(exit_code: int = 0, reason: str = 'normal') → None

Mark node as terminated with optional exit details

Parameters:

• exit_code (int) – Exit code of the node process

• reason (str) – Reason for termination

record_timestamp(event: MetricsEvent | str) → None

Record timestamp for a given event in RAM

Parameters:

event (Union[MetricsEvent, str]) – Event to record either as MetricsEvent enum or legacy string

abstract save_metrics() → None

Save metrics using the appropriate storage method

update_info(node_name: str, instance_name: str, transports: List[str] | None = None) → None

Update node and instance names after argument parsing

Parameters:

• node_name (str) – Name of the node

• instance_name (str) – Name of the instance

• transports (List[str], optional) – List of active transports

class gd_node.performance_metrics.collector.MetricsEvent(value)

Bases: Enum

Events that can be recorded during GD_Node lifecycle

IMPORTS_DONE = 'imports_done'

INIT_DONE = 'init_done'

INIT_START = 'init_start'

RUNTIME_START = 'runtime_start'

START = 'start'

TERMINATE_DONE = 'terminate_done'

TERMINATE_START = 'terminate_start'

class gd_node.performance_metrics.collector.NodeState(value)

Bases: Enum

States of a GD_Node during its lifecycle

RUNNING = 'running'

STARTING = 'starting'

TERMINATED = 'terminated'

class gd_node.performance_metrics.collector.NullMetricsCollector(*args: Any, **kwargs: Any)

Bases: MetricsCollector

No-op metrics collector when metrics are disabled

calculate_durations(*args: Any, **kwargs: Any) → Dict[str, float]

No-op for NullMetricsCollector

record_timestamp(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector

save_metrics(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector

update_info(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector

Module contents

Copyright (C) Mov.ai - All Rights Reserved Unauthorized copying of this file, via any medium is strictly prohibited Proprietary and confidential

class gd_node.performance_metrics.MetricsCollector(node_name: str | None = None, instance_name: str | None = None)

Bases: ABC

Collects and stores performance metrics for GD_Node execution. Time measurements are stored in seconds throughout the system. Only collects metrics if PLATFORM_METRICS environment variable is set.

calculate_durations() → Dict[str, float]

Calculate time durations from stored timestamps. These values are stored as floats to maintain precision.

Returns:

Dictionary of duration measurements in seconds

Return type:

Dict[str, float]

static create(*args: Any, **kwargs: Any) → NullMetricsCollector | InfluxDBMetricsCollector

Factory method that returns appropriate collector based on environment

mark_running() → None

Mark node as running after initialization is complete

mark_terminated(exit_code: int = 0, reason: str = 'normal') → None

Mark node as terminated with optional exit details

Parameters:

• exit_code (int) – Exit code of the node process

• reason (str) – Reason for termination

record_timestamp(event: MetricsEvent | str) → None

Record timestamp for a given event in RAM

Parameters:

event (Union[MetricsEvent, str]) – Event to record either as MetricsEvent enum or legacy string

abstract save_metrics() → None

Save metrics using the appropriate storage method

update_info(node_name: str, instance_name: str, transports: List[str] | None = None) → None

Update node and instance names after argument parsing

Parameters:

• node_name (str) – Name of the node

• instance_name (str) – Name of the instance

• transports (List[str], optional) – List of active transports

class gd_node.performance_metrics.MetricsEvent(value)

Bases: Enum

Events that can be recorded during GD_Node lifecycle

IMPORTS_DONE = 'imports_done'

INIT_DONE = 'init_done'

INIT_START = 'init_start'

RUNTIME_START = 'runtime_start'

START = 'start'

TERMINATE_DONE = 'terminate_done'

TERMINATE_START = 'terminate_start'

class gd_node.performance_metrics.NullMetricsCollector(*args: Any, **kwargs: Any)

Bases: MetricsCollector

No-op metrics collector when metrics are disabled

calculate_durations(*args: Any, **kwargs: Any) → Dict[str, float]

No-op for NullMetricsCollector

record_timestamp(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector

save_metrics(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector

update_info(*args: Any, **kwargs: Any) → None

No-op for NullMetricsCollector