[hardikkaurani]

Description

Fixes defects 1 and 2 of #5890. (Note: defect 3 is excluded as it is actively being addressed in open PR #5778).

1. MeterProvider.add_metric_reader() now calls metric_reader._set_meter_provider(self):
   Previously, add_metric_reader() called metric_reader._set_collect_callback(...) but never called metric_reader._set_meter_provider(self), leaving metric_reader._metrics bound to NoOpMeterProvider() and failing to produce reader collection duration self-observability metrics (otel.sdk.metric_reader.collection.duration).

2. Allocates MetricReader component ID once per instance:
   Previously, MetricReader.__init__ called create_metric_reader_metrics(..., NoOpMeterProvider(), ...), which prematurely incremented _component_counter[component_type] from 0 to 1 with a no-op provider. When MeterProvider subsequently called metric_reader._set_meter_provider(self), create_metric_reader_metrics was called a second time, incrementing the counter to 2 and naming the first instance <type>/1.
   Now, MetricReader.__init__ initializes self._metrics as NoOpMetricReaderMetrics(), create_metric_reader_metrics returns NoOpMetricReaderMetrics() when passed a NoOpMeterProvider, and MetricReader allocates and stores self._otel_component_id once upon registration (<type>/0 for the first instance).

Related Issues

• Fixes part of MetricReader self-observability: add_metric_reader() does not wire MeterProvider, collect() omits error.type on failure, and MeterProvider.__init__ double-increments component ID #5890 (defects 1 & 2)
• Excludes defect 3 to avoid overlapping with open PR Record error.type for failed metric reader collections #5778

Type of Change

• Bug fix (non-breaking change which fixes an issue)

[Copilot]

Copilot was unable to review this pull request because the user who requested the review has insufficient remaining quota. Try requesting a review at a lower effort level.

[opentelemetry-pr-dashboard]

Pull request dashboard status

Waiting on reviewers · refreshed 2026-11-08 04:00 UTC

Review the latest changes.

Status above doesn't look right?

• Just replied or pushed? Anything around or after the refresh time above may not be picked up yet — give it a few minutes.
• Anything look wrong? Report it with what you expected; it helps us improve the dashboard.

[hardikkaurani]

Hi @herin049, could you please take a look at this PR when you have some time?

This addresses defects 1 and 2 from #5890:

• Calls metric_reader._set_meter_provider(self) in MeterProvider.add_metric_reader(), ensuring readers registered dynamically have their self-observability collection duration metrics wired to the active provider.
• Allocates MetricReader component ID once per instance by returning NoOpMetricReaderMetrics when initialized with NoOpMeterProvider, preventing premature double-increments of _component_counter and correctly indexing the first instance as <type>/0.

(Note: defect 3 regarding error.type was excluded here as it is actively being addressed in open PR #5778).

Regression tests (with proper cleanup isolation) and Towncrier changelog fragment are included, and EasyCLA is signed.

Thanks!

[knowprasad]

Took a look through the changes in opentelemetry-sdk.

Calling metric_reader._set_meter_provider(self) directly in MeterProvider.add_metric_reader() makes sense and brings dynamically added readers to parity with readers supplied at provider initialization.

Initializing self._metrics as NoOpMetricReaderMetrics() and having create_metric_reader_metrics return early when passed NoOpMeterProvider cleanly avoids the double-increment bug on _component_counter. Allocating self._otel_component_id once and reusing it across re-bindings keeps the naming predictable (/0 for the first instance).

The regression tests in est_in_memory_metric_reader.py and est_periodic_exporting_metric_reader.py verify both the initial state and dynamic registration cleanly, and using self.addCleanup for reader and provider shutdowns ensures tests stay properly isolated.

Looks solid to me!

[hardikkaurani]

Thanks for taking the time to go through this and for the detailed feedback!

Glad the approach and the tests look good. I wanted to make sure dynamically added readers behave consistently with readers passed during provider initialization, without introducing any extra component ID increments.

Appreciate you taking a look! @knowprasad