[receiver/prometheusreceiver] Preserve bucketless histograms during NHCB conversion

[shubhgupta2104]

Assisted-by Github-Copilot

Description

Bucketless classic histograms contain _sum and _count series but no _bucket series.

When convert_classic_histograms_to_nhcb: true is enabled and classic histogram retention is disabled, Prometheus converts these histograms into an NHCB with empty CustomValues and a single implicit +Inf bucket. The receiver currently returns early for this representation, silently dropping the histogram.

This change preserves the implicit +Inf bucket, consistent with the behaviour established in #22070 and #23448.

There are no configuration or public API changes.

Link to tracking issue

Fixes #49893

Related to #22070, #23448 , and #43095 .

Testing

Added receiver tests covering:

• Integer histogram values
• Floating-point histogram values
• Stale histogram samples
• Complete receiver scrape behavior

Documentation

Added a user-facing changelog entry describing the fix.

Authorship

• I, a human, wrote this pull request description myself.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: shubhgupta2104 / name: Shubh Gupta (1dc8722)

[github-actions]

Welcome, contributor! Thank you for your contribution to opentelemetry-collector-contrib.

Important reminders:

• Read our Contributing Guidelines.
• Sign the CLA if you haven't already.
• Follow the OpenTelemetry Generative AI policy: disclose any AI use in your contribution, and communicate (PR descriptions, review replies) in your own words rather than AI-generated text.
• Give reviewers at least a few days before pinging them for feedback.
• If you need help with general setup, development process, or contributor etiquette, ask in #opentelemetry-new-contributors.
• First-time contributors should have at most one PR not marked as draft until their first PR is merged.
• If your change isn't one of our priority components, reviews may take more time.
• Raise technical or Collector-specific questions in #otel-collector-dev or a Collector SIG meeting.

[opentelemetry-pr-dashboard]

Pull request dashboard status

Waiting on the author · refreshed 2026-08-02 18:42 UTC

Respond to 1 review item (e.g. link a commit, explain why not, ask a follow-up):

• Top-level threads: 1

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.
• Should this be with reviewers? Comment /dashboard route:reviewers to route it to them.
• Anything wrong — including the routing? Report it with what you expected; it helps us improve the dashboard.

[shubhgupta2104]

I found in a follow-up review that the fix also starts emitting semantically inconsistent _sum-only histogram families.

For example, a histogram containing _sum=123.5 but no _count can now become:

count=0
sum=123.5
buckets [0]

That is semantically questionable and differs from the classic path, which drops it. Prometheus’s conversion loses whether _count was absent, so fixing this cleanly may require more than checking Count == 0

Can you help me understand:

• Should the receiver faithfully preserve Prometheus’s converted NHCB?
• Or should it reject the semantically inconsistent count=0, sum!=0 case?

[dashpole]

Looks correct to me. @krajorama should take a look as well

[Copilot]

Pull request overview

Fixes a regression in the Prometheus receiver’s classic-histogram → NHCB conversion path where “bucketless” classic histograms (only _sum/_count, no _bucket series) could be silently dropped. The change aligns NHCB conversion behavior with the existing classic-histogram handling established in prior work (#22070 / #23448), preserving the implicit +Inf bucket representation instead of returning early.

Changes:

• Stop treating NHCB histograms with empty CustomValues (no explicit bounds) as “unsupported” in toDistributionPoint, so bucketless histograms are still emitted.
• Add unit and end-to-end receiver tests covering integer/float NHCB bucketless histograms, including stale samples and scrape behavior when classic retention is toggled.
• Add a user-facing changelog entry describing the bug fix.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated no comments.

File	Description
receiver/prometheusreceiver/internal/metricfamily.go	Removes early-return behavior that dropped NHCB bucketless histograms during conversion to OTLP histograms.
receiver/prometheusreceiver/internal/metricfamily_test.go	Adds NHCB unit tests for empty explicit bounds (bucketless) for both integer and float histograms, including stale cases.
receiver/prometheusreceiver/metrics_receiver_test.go	Adds an end-to-end scrape test ensuring bucketless classic histograms are preserved when converted to NHCB (with/without classic retention).
.chloggen/fix-prometheusreceiver-bucketless-nhcb.yaml	Adds a changelog entry documenting the fix for users.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[shubhgupta2104]

@dashpole I think you will need to click Approve and run workflow again, because I added a merge commit. Also, any thoughts on my comment above?

[singhvibhanshu]

/workflow-approve

[dashpole]

• Should the receiver faithfully preserve Prometheus’s converted NHCB?
• Or should it reject the semantically inconsistent count=0, sum!=0 case?

It should be consistent with the behavior in the classic path. So if it is rejected in that case, i'm OK with rejecting it here as well.

[singhvibhanshu]

/workflow-approve

[shubhgupta2104]

@krajorama Can you also take a look please? Thanks :)