Why python prometheus client collectors create metric object every time when collect method is invoked

Question

https://github.com/prometheus/client_python/blob/master/prometheus_client/gc_collector.py

import gc
import platform
from typing import Iterable

from .metrics_core import CounterMetricFamily, Metric
from .registry import Collector, CollectorRegistry, REGISTRY


class GCCollector(Collector):
    """Collector for Garbage collection statistics."""

    def __init__(self, registry: CollectorRegistry = REGISTRY):
        if not hasattr(gc, 'get_stats') or platform.python_implementation() != 'CPython':
            return
        registry.register(self)

    def collect(self) -> Iterable[Metric]:
        collected = CounterMetricFamily(
            'python_gc_objects_collected',
            'Objects collected during gc',
            labels=['generation'],
        )
        uncollectable = CounterMetricFamily(
            'python_gc_objects_uncollectable',
            'Uncollectable objects found during GC',
            labels=['generation'],
        )

        collections = CounterMetricFamily(
            'python_gc_collections',
            'Number of times this generation was collected',
            labels=['generation'],
        )

        for gen, stat in enumerate(gc.get_stats()):
            generation = str(gen)
            collected.add_metric([generation], value=stat['collected'])
            uncollectable.add_metric([generation], value=stat['uncollectable'])
            collections.add_metric([generation], value=stat['collections'])

        return [collected, uncollectable, collections]

Instead of re-using metric object it creates it every time just to log one value. What if I need to pull metric every second? does it mean it will create it every second? it's not an optimal at all. I feel that under certain circumstances it may contribute into memory allocation even more than time app

What I'm missing? Is it by design?

Shivam Bhosle · Accepted Answer · 2025-07-18 11:00:58Z

0

Yes, this is by design. It's by design for custom collectors for reasons like Accuracy , Simplicity , Low Impact:

This pattern is for custom collectors pulling external data. Standard metrics like Counter are stateful and long-lived.

answered Jul 18 at 11:00

Shivam Bhosle

4531 silver badge6 bronze badges

Sign up to request clarification or add additional context in comments.

1 Comment

slesh Jul 21 at 20:42

> for reasons like Accuracy , Simplicity , Low Impact I didn't get this. What does it mean? How to use those custom collectors if it allocates lots of object whenever I send metrics out

Collectives™ on Stack Overflow

Why python prometheus client collectors create metric object every time when collect method is invoked

1 Answer 1

1 Comment

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Related