Claude Issue Review

[ocelotl]

I asked Claude to review every open issue we have and this is what it produced:

opentelemetry-python issue triage: 358 open issues, sorted by what to keep

Reference release: 1.43.0 / 0.64b0
Method: Full open-issue list pulled via gh issue list, clustered by title/label similarity, then each candidate verified individually against issue bodies, linked PRs, current source, and CHANGELOG.md before being included below.

Summary

Metric	Count
Issues reviewed	358
Recommended for closing	~38
Confirmed not duplicates	9
Worth a second look	~17

Duplicate clusters

High confidence

B3 propagator crashes on partial headers

• Clients requests with partial set of B3 propagation headers causes crash #2629 — canonical, 2022-04-20
• Documentation request - python opentelemetry-propagator-b3 #2667 — close

The reporter of #2667 explicitly ties their crash to #2629 in the issue body and never followed up afterward.

Prometheus exporter multiprocess support

• Prometheus exporter does not work with multiprocessing #3747 — canonical, 2024-03-02
• Migrating from prometheus client - MultiProcessCollector feature #4524 — close

A maintainer comment on #4524 already links it back to #3747.

Context detach errors on suspended generators

• Runtime context fails to detach token #2606 — canonical, 2022-04-15, 38 comments
• Logged exception "Failed to detach context" when testing Flask app #3919 — close

#2606 already has root-cause analysis from the original investigation. #3919 is a Flask-specific repro of the exact same generator-suspension bug.

OTLP exporter partial-success handling

• Handle partial success responses from OTLP export services #2882 — 2022-08-19, 0 comments — close
• Make it possible to opt-in to logging partial errors #4803 — canonical, 2025-11-10

#2882 sat with zero engagement for three years; #4803 restates the same gap with a more actionable description — keep that one instead.

Worth a second look

Fork-safety reports
#3307 (umbrella), #4994, #4215, #4759

#4994 and #4215 describe distinct fork-safety bugs and should stay open under the #3307 umbrella. #4759 looks like a near-duplicate of #3307 itself — reasonable to close.

Metrics break after os.fork()
#2767, #3402

These two look like genuine duplicates of each other — worth a maintainer call on which to keep rather than an automated one.

B3 sampled-flag handling — not duplicates
#2341, #2629, #4793

Similar titles, but each describes a distinct edge case in B3 sampled-flag parsing — keep all three separate.

Exporter timeout handling
#1860, #4555 (canonical), #346

Fold #1860 into #4555 as the same report. #346 is a long-standing, broader tracking issue — keep it open separately.

Declarative file configuration sub-tasks — confirmed not duplicates
#5073, #5126, #5229, #5318, #5334, #5335, #5346, #5352, #5355

These look like one sprawling cluster by title but are confirmed sub-tasks of the active declarative-config epic, #3631. Leave all of them open.

Closeable as no-longer-relevant

Fix already shipped, issue never auto-closed

• OTEL_PYTHON_CONTEXT not working #3857 — PR api: conditionally load entrypoints for OTEL_PYTHON_CONTEXT #5144, in 1.43.0 CHANGELOG
• Wrong test value for default timeout in OTLP Exporter Mixin #4057 — fixed by PR Delete duplicated OTLP Exporter tests, move them to the mixin unit test. Fix broken shutdown unit test. #4504
• TestConcurrentMultiSpanProcessor.test_force_flush_late_by_timeout failure #4543 — fixed by PR fix(sdk): improve force flush logic #5179 / opentelemetry-sdk: make test_force_flush_late_by_timeout less flaky on pypy/windows #5102
• FIX: _SPAN_ID_MAX_VALUE should be generated within the size int64 #3535 — RandomIdGenerator fix shipped
• Update headers for benchmarks website #3591 — PR Update headers for benchmarks website #3592 merged Jan 2024
• Restore the correct link for published benchmarks #2820 — fix PR merged

Already implemented in current code

Follow-up research to attach a fixing PR to each of these turned up three
(#2951, #3032, #3097) with no merged PR at all — they may not actually
be resolved and are worth re-reading before closing.

Issue	Fixed by	Confidence	Note
#2941	PR #3265	High	maintainer commented on the issue directly linking this PR as the fix
#2951	none found	Low	closest is PR #3267, but it benchmarks instrument calls, not the aggregators this issue asked for — likely still open
#3032	none found	Low	exploratory design ask; appears genuinely unaddressed
#3097	none found	Low	a dedicated fix, PR #3107, was opened but closed unmerged in 2024
#3108	PR #3138	Low–medium	added non-blocking OTLP exporter shutdown, but doesn't cite this issue
#3144	PR #3251	Medium	semconv 1.21.0 update replaced net.* attributes; no explicit issue reference
#3145	PR #3251	Medium	same PR as #3144, also added the messaging.* semconv attributes
#3219	PR #3978	Low–medium	fixed exponential histogram merge bugs, but different specific bugs than this issue named
#3322	PR #2964, PR #3798	Medium	#2964 added exponential histogram min/max, #3798 fixed its collection correctness
#3336	PR #4361	Medium	implemented ExplicitBucketBoundaries directly as a stable API
#3637	PR #5131, #5095, #5098, #5128, #5129	Medium–high	declarative-config plugin-loading epic (sampler/propagator/exporter/resource-detector loading)
#3651	no single PR — added piecemeal, see below	High	the dedicated fix PR (#3662) was never merged
#2217	PR #2215	High	contributor comment on the issue confirms this PR as the fix
#2246	PR #2296	Medium–high	replaced the exact race-condition-prone registration methods this issue flagged
#2554	PR #2853	Medium–high	added force_flush() cascading to push metric exporters
#4438	PR #4676	Medium	added the -> LogExportResult return type as part of a larger refactor

__repr__ for #3651 was added file-by-file, independent of that issue, by:

• sdk/trace/sampling.py — PR Move samplers to SDK package #1023
• sdk/trace/__init__.py (Span) — PR Span creation in tracer SDK #69, tweaked by Named tracers #301; (SpanLimits) — PR Allow users to "unset" SDK limits and evaluate default limits lazily instead of on import #1839
• sdk/_logs/_internal/__init__.py — PR Use BoundedAttributes instead of raw dict to extract attributes from LogRecord #3114 #3310
• sdk/util/instrumentation.py (InstrumentationInfo) — PR Implement "named" meters + Remove "Batcher" from Meter constructor #431; (InstrumentationScope) — PR Add InstrumentationScope and deprecate InstrumentationLibraryInfo #2583, tweaked by Fix #3695: add attributes to get_meter fn and InstrumentationScope #4015
• api/trace/span.py (TraceState) — PR Fix TraceState to adhere to specs #1502; (SpanContext, NonRecordingSpan) — PR api: adding trace.get_current_span #552, tweaked by Add some repr()s to basic trace types. #1485
• api/metrics/_internal/observation.py — PR Make metrics public #2684

Dead or removed subsystems

• opentelemetry-exporter-jaeger-proto-grpc: export tests failed #1975 — Jaeger exporters fully removed
• Tests should use configured provider types for non-standard SDKs #486 — old ext/ tooling layout
• Add benchmarks to every type of package in build workflow #1468 — eachdist.py, no longer used
• eachdist.py script cannot install the test/utils package #1846
• Requirement already satisfied shows many times in logs #2087
• Remove eachdist #1462

Answered by a maintainer, silent 1+ year

• trace.get_current_span doesn't provide active span under test #3999
• OpenTelemetry Python Auto-Instrumentation Image (v0.47b) Not Compatible with Seldon Core (v1.17.1) Image #4189
• Opentelemetry python version compatibility issue #4241
• AssertionError in OtelBatchSpanProcessor #4025
• Support Opentelemetry Proto definitions for span attributes #4380
• Is it possible to use opentelemetry as downstream of an old system using zipkin? #4387
• High Latency When Using Nginx Ingress with OTel Collector over HTTP #4488
• Configuring Span Attributes for Http Request programmatically #3320
• Little confused about B3MultiFormat key name spell #3598
• Use Gitchangelog #2530

Wrong repo — already fixed in opentelemetry-python-contrib

• Asyncio : trace_to_thread raise exception when func doesn't has __name__ attribute #4207

Lower confidence — flagged but not fully verified

Old and quiet, but not confirmed against source/CHANGELOG. Read before closing.

• Make Exporters Shutdown spec-compliant #1791
• Handle Exceptions without Span #2403
• Debugging - InMemoryMetricReader: MetricsTimeoutError('Timed out while flushing metric readers') #4196
• C++/python interoperability #2691
• Auto-instrumentation honouring the OTEL_PYTHON_X_PROVIDER env-vars #2744
• Protect auto instrumentation against unintended usage #2073
• Implement error handling in tracing #2275
• Getting exception while using PeriodicExportingMetricReader from export library #3246
• Test that metrics methods can be called concurrently #2128
• Metric exporters should report and error condition... #2562
• Add all environment variable values defined in the spec to the environment variables modules #2994

Spot-checked and kept open

#3537 ("fix the cyclic imports") was checked against the current tree — the exact import workaround it flags is still present across six files in opentelemetry-sdk/.../metrics/_internal/, with no PR or CHANGELOG entry addressing it. Genuinely still open.

[ocelotl]

I have been manually reviewing the results above, so far here are the issues I suggest closing:

#2629
#2667
#4803
#4057
#4543
#3857
#3591
#3535
#2820
#2941
#3108
#3144
#3145
#3336
#3637
#3651
#2217
#2246
#2554
#4438
#1975
#486
#1468
#1846
#2087
#4189
#4241
#4380
#3320
#5373

[ocelotl]

@open-telemetry/python-approvers please assign this issue to me