OM2: WIP collab 03.2026 by bwplotka · Pull Request #2851 · prometheus/docs

bwplotka · 2026-02-25T15:10:34Z

WIP

Signed-off-by: bwplotka <bwplotka@gmail.com> tmp Signed-off-by: bwplotka <bwplotka@gmail.com> tmp Signed-off-by: bwplotka <bwplotka@gmail.com> update Signed-off-by: bwplotka <bwplotka@gmail.com>

bwplotka · 2026-03-20T12:23:07Z

Rebased again upfront our WG sync

Signed-off-by: bwplotka <bwplotka@gmail.com>

docs/specs/om/open_metrics_spec_2_0.md

Apply non-comment content changes from PR #2851: - StateSet: clarify "one Metric with a boolean value per state" and add "for a single Timestamp" to ENUM constraint - Info: remove sentence about encoding ENUMs (to be reconsidered) - Histogram: fix "Histogram Sample" -> "Histogram" in Sum definition - Histogram/GaugeHistogram: fix "the later" -> "the latter" (typo) - Histogram: simplify "A Histogram Sample SHOULD have a Start Timestamp" - Classic Buckets: fix "This allow" -> "This allows" and "performance/anti-denial-of-service" -> "performance or anti-denial-of-service" - Classic Buckets: spell out "one second" instead of "1 second" - Native Buckets: "8 bit" -> "8-bit" - Native Buckets: remove reserved schema value details (-9..-5, 9..52, -53) - Native Buckets: use backticks for n, i, i+1 inline code - Native Buckets: remove "in the following" from MaxFloat64/MinFloat64 parenthetical - Native Buckets: fix "an lower" -> "a lower" - Summary: update description to mention "precomputed quantiles" instead of "average event size" - Summary: move SHOULD NOT sentence before backwards compatibility sentence - Summary: split Start Timestamp sentence and change "relate to" -> "be based on" - Summary: add backticks around metric name example - Protocol Negotiation: capitalize "Ingestor" mid-sentence - ABNF: remove HTML editor note comment Signed-off-by: György Krajcsovits <gyorgy.krajcsovits@grafana.com> Coded with Claude Sonnet 4.6.

docs/specs/om/open_metrics_spec_2_0.md

krajorama · 2026-03-26T07:44:25Z

docs/specs/om/open_metrics_spec_2_0.md

 ##### Number

+// MAYBE: Clarify floating point exactly / link to where we do this instead of float64
+// MAYBE: "ut it MAY be used to signal a division by zero" -> add or any result of the math operations that would result it in.


both in #2928

docs/specs/om/open_metrics_spec_2_0.md

krajorama · 2026-03-26T09:20:37Z

docs/specs/om/open_metrics_spec_2_0.md

+// MAYBE: Require encryption? 1.2 is safe? 
+All ingestor implementations MUST be able to ingest data secured with TLS 1.2 or later. All exposers SHOULD be able to emit data secured with TLS 1.2 or later. Ingestor implementations SHOULD be able to ingest data from HTTP without TLS. All implementations SHOULD use TLS to transmit data.

+// TODO: Fix the sentence 


Not sure what we meant here. Out-of-band means to me that the negotiation isn't part of the OM2.0 text format, which it isn't so that tracks.

For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

This is not really english, still TODO

docs/specs/om/open_metrics_spec_2_0.md

krajorama · 2026-03-26T11:56:22Z

docs/specs/om/open_metrics_spec_2_0.md


 The OpenMetrics formats are Regular Chomsky Grammars, making writing quick and small parsers possible.

+// MAYBE: Be clear on failure modes.


asked claude:
Two relevant statements, and they are consistent:

Line 380 (Text Format section): "Partial or invalid expositions MUST be considered erroneous in their entirety." — the whole exposition is rejected, not just the affected metric.

Line 1405 (Size Limits section): "If any single value/metric/exposition exceeds such limits then the whole exposition must be rejected." — same conclusion for size limit violations.

The only exception is line 163: "Ingestors MAY reject such MetricFamily" — which allows per-MetricFamily rejection for discouraged name suffixes, but this is a MAY and is specific to that one case.

So the general rule is whole-exposition rejection, with a narrow carve-out for discouraged suffixes.

Signed-off-by: bwplotka <bwplotka@gmail.com>

docs/specs/om/open_metrics_spec_2_0.md

ywwg · 2026-03-30T14:05:40Z

docs/specs/om/open_metrics_spec_2_0.md

 All ingestor implementations MUST be able to ingest data secured with TLS 1.2 or later. All exposers SHOULD be able to emit data secured with TLS 1.2 or later. Ingestor implementations SHOULD be able to ingest data from HTTP without TLS. All implementations SHOULD use TLS to transmit data.

+// TODO: Fix the sentence 
 Negotiation of what version of the OpenMetrics format to use is out-of-band. For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.


Suggested change

Negotiation of what version of the OpenMetrics format to use is out-of-band. For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

Negotiation of which version of the OpenMetrics format to use is out-of-band. For example, for HTTP pull-based exposition, standard HTTP content type negotiation is used and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

bwplotka · 2026-03-30T13:58:51Z

docs/specs/om/open_metrics_spec_2_0.md


 Info metrics are used to expose textual information which SHOULD NOT change during process lifetime. Common examples are an application's version, revision control commit, and the version of a compiler.

+// CONSISTENCY: Last pass of Name/name (definition) 


bwplotka · 2026-03-30T14:03:13Z

docs/specs/om/open_metrics_spec_2_0.md

+<!---
+// TODO(??) give example of unknown metadata? No meta exposed, two differently name metric families.
+-->


Suggested change

bwplotka · 2026-03-30T14:03:44Z

docs/specs/om/open_metrics_spec_2_0.md


 Info metrics are used to expose textual information which SHOULD NOT change during process lifetime. Common examples are an application's version, revision control commit, and the version of a compiler.

+// CONSISTENCY: Last pass of Name/name (definition) 


Suggested change

// CONSISTENCY: Last pass of Name/name (definition)

bwplotka · 2026-03-30T14:04:44Z

docs/specs/om/open_metrics_spec_2_0.md

 All ingestor implementations MUST be able to ingest data secured with TLS 1.2 or later. All exposers SHOULD be able to emit data secured with TLS 1.2 or later. Ingestor implementations SHOULD be able to ingest data from HTTP without TLS. All implementations SHOULD use TLS to transmit data.

+// TODO: Fix the sentence 
 Negotiation of what version of the OpenMetrics format to use is out-of-band. For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.


Suggested change

Negotiation of what version of the OpenMetrics format to use is out-of-band. For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

Negotiation of what version of the OpenMetrics format to use is out-of-band. For example, for pull-based exposition over HTTP standard, when HTTP content type negotiation is used, it MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

bwplotka · 2026-03-30T14:05:18Z

docs/specs/om/open_metrics_spec_2_0.md

+// TODO: Fix the sentence 
 Negotiation of what version of the OpenMetrics format to use is out-of-band. For example for pull-based exposition over HTTP standard HTTP content type negotiation is used, and MUST default to the oldest version of the standard (i.e. 1.0.0) if no newer version is requested.

+// MAYBE: Exposer? Also fallback to text format?


Suggested change

// MAYBE: Exposer? Also fallback to text format?

bwplotka force-pushed the bwplotka-omwg-review branch from 8f1bf07 to 9ada8a5 Compare March 4, 2026 08:10

bwplotka force-pushed the bwplotka-omwg-review branch from 9ada8a5 to 277398e Compare March 13, 2026 11:35

bwplotka changed the title ~~[WIP] OM2: WG changes for 2.0-rc.0~~ OM2: WG agreed changes for 2.0-rc.0 Mar 13, 2026

bwplotka changed the base branch from main to spec-fmt March 13, 2026 11:36

bwplotka marked this pull request as ready for review March 13, 2026 11:36

bwplotka mentioned this pull request Mar 13, 2026

- #2893

Closed

bwplotka force-pushed the spec-fmt branch from 128f74e to c2c3aa3 Compare March 13, 2026 12:35

Base automatically changed from spec-fmt to main March 13, 2026 13:04

bwplotka changed the title ~~OM2: WG agreed changes for 2.0-rc.0~~ OM2: WIP collab 13.03.2026 Mar 13, 2026

krajorama mentioned this pull request Mar 16, 2026

OM 2.0: Remove MetricPoint concept, and use Sample #2894

Merged

bwplotka mentioned this pull request Mar 17, 2026

om2: pre rc.0 changes from WG #2895

Merged

bwplotka force-pushed the bwplotka-omwg-review branch from acd01d9 to 74bd60e Compare March 17, 2026 13:01

bwplotka changed the base branch from main to om2postreviewpart1 March 17, 2026 13:01

bwplotka changed the base branch from om2postreviewpart1 to main March 17, 2026 13:07

bwplotka force-pushed the bwplotka-omwg-review branch from aded6ca to 0badf83 Compare March 17, 2026 20:46

bwplotka changed the title ~~OM2: WIP collab 13.03.2026~~ OM2: WIP collab 03.2026 Mar 17, 2026

om2: pre rc.0 changes from WG

6317f09

Signed-off-by: bwplotka <bwplotka@gmail.com> tmp Signed-off-by: bwplotka <bwplotka@gmail.com> tmp Signed-off-by: bwplotka <bwplotka@gmail.com> update Signed-off-by: bwplotka <bwplotka@gmail.com>

bwplotka force-pushed the bwplotka-omwg-review branch from 0badf83 to 6317f09 Compare March 20, 2026 12:22

bwplotka added 2 commits March 20, 2026 13:26

Merge branch 'main' into bwplotka-omwg-review

684ec09

update

c940d03

Signed-off-by: bwplotka <bwplotka@gmail.com>

krajorama reviewed Mar 25, 2026

View reviewed changes