promql: reuse `histogramStatsIterator` where possible, and expose it for other implementations to use #16686

charleskorn · 2025-06-04T04:16:42Z

This PR builds upon #14097.

It makes two changes:

It exposes the histogramStatsIterator type so that other query engine implementations can use it.
It adds a Reset method that allows the iterator to be reused, reducing allocations and improving performance.

The existing benchmarks show a 2-4% reduction in allocations:

goos: darwin
goarch: arm64
pkg: github.com/prometheus/prometheus/promql
cpu: Apple M1 Pro
                                                                          │ before.txt  │             after.txt             │
                                                                          │   sec/op    │   sec/op     vs base              │
NativeHistograms/histogram_count_with_short_rate_interval-10                425.7m ± 1%   423.5m ± 1%       ~ (p=0.065 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                 721.4m ± 1%   715.2m ± 2%       ~ (p=0.093 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10   388.5m ± 3%   391.4m ± 2%       ~ (p=0.132 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10    683.9m ± 1%   680.1m ± 2%       ~ (p=0.093 n=6)

                                                                          │  before.txt   │              after.txt              │
                                                                          │     B/op      │     B/op       vs base              │
NativeHistograms/histogram_count_with_short_rate_interval-10                151.2Mi ±  0%   146.8Mi ±  0%  -2.92% (p=0.002 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                 148.0Mi ±  4%   143.5Mi ±  4%  -2.98% (p=0.041 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10   146.9Mi ± 10%   143.5Mi ± 10%       ~ (p=0.180 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10    157.8Mi ±  0%   154.4Mi ±  0%  -2.14% (p=0.002 n=6)

                                                                          │ before.txt  │             after.txt             │
                                                                          │  allocs/op  │  allocs/op   vs base              │
NativeHistograms/histogram_count_with_short_rate_interval-10                1.792M ± 0%   1.726M ± 0%  -3.68% (p=0.002 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                 1.684M ± 3%   1.618M ± 3%  -3.92% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10   1.690M ± 2%   1.651M ± 2%  -2.31% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10    1.611M ± 0%   1.572M ± 0%  -2.42% (p=0.002 n=6)

The existing benchmarks don't show a noticeable improvement in latency as the benchmarks run a range query with 240 steps over 3000 series, and so the benchmark time is dominated by other things.

Modifying the existing benchmarks to run an instant query shows more dramatic results:

goos: darwin
goarch: arm64
pkg: github.com/prometheus/prometheus/promql
cpu: Apple M1 Pro
                                                                          │ before-instant.txt │         after-instant.txt          │
                                                                          │       sec/op       │   sec/op     vs base               │
NativeHistograms/histogram_count_with_short_rate_interval-10                       23.22m ± 1%   20.74m ± 1%  -10.67% (p=0.002 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                        64.70m ± 1%   62.16m ± 1%   -3.94% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10          15.41m ± 1%   13.53m ± 3%  -12.17% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10           51.41m ± 1%   49.14m ± 3%   -4.41% (p=0.002 n=6)

                                                                          │ before-instant.txt │          after-instant.txt          │
                                                                          │        B/op        │     B/op      vs base               │
NativeHistograms/histogram_count_with_short_rate_interval-10                      7.538Mi ± 0%   2.112Mi ± 0%  -71.98% (p=0.002 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                       9.353Mi ± 0%   3.958Mi ± 1%  -57.68% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10         6.981Mi ± 0%   2.390Mi ± 0%  -65.77% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10          8.801Mi ± 0%   4.222Mi ± 1%  -52.02% (p=0.002 n=6)

                                                                          │ before-instant.txt │         after-instant.txt          │
                                                                          │     allocs/op      │  allocs/op   vs base               │
NativeHistograms/histogram_count_with_short_rate_interval-10                      126.25k ± 0%   36.27k ± 0%  -71.27% (p=0.002 n=6)
NativeHistograms/histogram_count_with_long_rate_interval-10                       171.35k ± 0%   81.38k ± 0%  -52.51% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_short_rate_interval-10          99.24k ± 0%   42.26k ± 0%  -57.42% (p=0.002 n=6)
NativeHistogramsCustomBuckets/histogram_count_with_long_rate_interval-10          138.35k ± 0%   81.36k ± 0%  -41.19% (p=0.002 n=6)

Signed-off-by: Charles Korn <charles.korn@grafana.com>

beorn7 · 2025-06-04T11:56:34Z

@fpetkovski might be interested in this… (and the rest of us might value his verdict ;).

beorn7

Generally looks good, but you need to pull in the current main with #16682 merged, resolve the conflicts, and then check if things still work. (As said in the comments, I have a few suspicions.)

promql/histogram_stats_iterator.go

# Conflicts: # promql/histogram_stats_iterator.go

Signed-off-by: Charles Korn <charles.korn@grafana.com>

#### What this PR does This PR implements an optimisation inspired by prometheus/prometheus#14097. I've implemented this as a MQE optimisation pass (rather than reusing the logic from that Prometheus PR) for two reasons: * It helps prove out the structure of optimisation passes, and demonstrate they're applicable to more than just common subexpression elimination. * We need to consider how this optimisation interacts with common subexpression elimination. Specifically, we need to consider all the different paths to a selector after common subexpression elimination has been performed to know if we can skip decoding histogram buckets. Alternatively, if we were to reuse the logic from Prometheus, the common subexpression elimination pass would need to know how to merge selectors that are identical save for the `SkipHistogramBuckets` hint. Peak memory consumption is reduced by ~10% in our benchmarks. Latency improves by up to 25% in some cases, but regresses by a small amount (in both absolute and relative terms) for some instant queries: ``` goos: darwin goarch: arm64 pkg: github.com/grafana/mimir/pkg/streamingpromql/benchmarks cpu: Apple M1 Pro │ original.txt │ final.txt │ │ sec/op │ sec/op vs base │ Query/histogram_count(sum(rate(nh_1[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 244.8µ ± 4% 236.4µ ± 1% -3.40% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 291.7µ ± 4% 261.4µ ± 2% -10.40% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 791.2µ ± 1% 627.4µ ± 2% -20.70% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 7.719m ± 1% 7.948m ± 0% +2.97% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 11.85m ± 1% 10.64m ± 1% -10.25% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 54.10m ± 3% 41.61m ± 2% -23.08% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 146.2m ± 5% 150.3m ± 0% ~ (p=0.065 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 223.9m ± 3% 202.3m ± 0% -9.63% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1047.5m ± 3% 789.9m ± 1% -24.60% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 264.6µ ± 4% 243.7µ ± 1% -7.92% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 359.7µ ± 4% 320.3µ ± 2% -10.96% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1.295m ± 3% 1.089m ± 0% -15.92% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 8.048m ± 1% 8.472m ± 1% +5.26% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 16.53m ± 2% 15.48m ± 4% -6.34% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 100.81m ± 1% 84.79m ± 0% -15.89% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 151.2m ± 0% 160.0m ± 0% +5.83% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 314.8m ± 0% 295.9m ± 1% -6.00% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1.968 ± 1% 1.660 ± 0% -15.64% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 244.1µ ± 2% 230.4µ ± 2% -5.60% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 289.1µ ± 2% 261.0µ ± 1% -9.70% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 783.2µ ± 1% 623.1µ ± 0% -20.44% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 7.820m ± 2% 7.937m ± 1% +1.49% (p=0.041 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 11.95m ± 1% 10.57m ± 1% -11.52% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 54.67m ± 1% 41.65m ± 0% -23.83% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 146.8m ± 1% 150.1m ± 0% +2.27% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 224.9m ± 0% 200.4m ± 1% -10.90% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1050.1m ± 1% 790.5m ± 1% -24.72% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 258.7µ ± 2% 243.4µ ± 2% -5.91% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 353.7µ ± 3% 317.4µ ± 2% -10.27% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1.290m ± 1% 1.084m ± 1% -15.93% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 8.085m ± 0% 8.471m ± 1% +4.78% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 16.58m ± 6% 15.49m ± 1% -6.59% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 101.78m ± 3% 84.98m ± 2% -16.50% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 152.6m ± 2% 160.2m ± 2% +4.99% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 317.7m ± 1% 296.4m ± 1% -6.70% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 1.980 ± 1% 1.654 ± 0% -16.49% (p=0.002 n=6) geomean 15.01m 13.55m -9.76% │ original.txt │ final.txt │ │ B │ B vs base │ Query/histogram_count(sum(rate(nh_1[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 76.69Mi ± 5% 68.97Mi ± 5% -10.07% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 74.16Mi ± 2% 66.49Mi ± 1% -10.34% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 73.48Mi ± 1% 64.20Mi ± 2% -12.63% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 71.55Mi ± 1% 62.59Mi ± 3% -12.52% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 69.62Mi ± 1% 62.04Mi ± 2% -10.89% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 71.49Mi ± 1% 63.12Mi ± 1% -11.71% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 75.71Mi ± 2% 67.80Mi ± 2% -10.44% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 75.67Mi ± 2% 67.84Mi ± 2% -10.36% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 79.64Mi ± 1% 72.62Mi ± 1% -8.81% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 75.88Mi ± 2% 67.17Mi ± 2% -11.48% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 73.75Mi ± 2% 66.02Mi ± 4% -10.49% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_1[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 73.00Mi ± 1% 64.12Mi ± 2% -12.16% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 70.35Mi ± 2% 63.04Mi ± 2% -10.39% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 69.81Mi ± 1% 61.73Mi ± 1% -11.57% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_100[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 71.08Mi ± 1% 62.70Mi ± 1% -11.78% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 74.77Mi ± 3% 67.80Mi ± 3% -9.32% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 75.47Mi ± 1% 66.90Mi ± 2% -11.36% (p=0.002 n=6) Query/histogram_count(sum(rate(nh_2000[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 79.35Mi ± 1% 70.52Mi ± 2% -11.14% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 76.41Mi ± 2% 69.45Mi ± 4% -9.10% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 75.35Mi ± 3% 66.16Mi ± 2% -12.19% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 72.92Mi ± 1% 63.72Mi ± 1% -12.62% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 71.88Mi ± 2% 62.25Mi ± 1% -13.40% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 70.02Mi ± 1% 61.74Mi ± 1% -11.83% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 71.70Mi ± 1% 63.26Mi ± 1% -11.78% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_instant_query/engine=MimirWithQueryPlanner-10 74.36Mi ± 2% 66.23Mi ± 2% -10.94% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 75.64Mi ± 1% 67.31Mi ± 1% -11.01% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[2m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 79.44Mi ± 1% 72.38Mi ± 1% -8.89% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 75.46Mi ± 4% 67.14Mi ± 2% -11.03% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 73.41Mi ± 3% 65.93Mi ± 3% -10.19% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_1[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 72.95Mi ± 1% 63.78Mi ± 1% -12.57% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 71.73Mi ± 2% 63.17Mi ± 2% -11.93% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 69.86Mi ± 1% 61.66Mi ± 1% -11.73% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_100[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 71.34Mi ± 1% 62.35Mi ± 2% -12.60% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_instant_query/engine=MimirWithQueryPlanner-10 75.13Mi ± 1% 67.05Mi ± 3% -10.75% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_range_query_with_100_steps/engine=MimirWithQueryPlanner-10 75.03Mi ± 1% 67.27Mi ± 2% -10.35% (p=0.002 n=6) Query/histogram_sum(sum(rate(nh_2000[20m]))),_range_query_with_1000_steps/engine=MimirWithQueryPlanner-10 78.70Mi ± 2% 70.84Mi ± 1% -9.98% (p=0.002 n=6) geomean 73.92Mi 65.69Mi -11.13% ``` ⚠️ This PR is pending prometheus/prometheus#16686 being merged and vendored into Mimir. #### Which issue(s) this PR fixes or relates to #10067 #### Checklist - [x] Tests updated. - [n/a] Documentation added. - [covered by #10067] `CHANGELOG.md` updated - the order of entries should be `[CHANGE]`, `[FEATURE]`, `[ENHANCEMENT]`, `[BUGFIX]`. - [n/a] [`about-versioning.md`](https://github.com/grafana/mimir/blob/main/docs/sources/mimir/configure/about-versioning.md) updated with experimental features.

charleskorn added 2 commits June 4, 2025 13:43

Expose type

b979467

Signed-off-by: Charles Korn <charles.korn@grafana.com>

Add Reset method

ce49220

Signed-off-by: Charles Korn <charles.korn@grafana.com>

charleskorn mentioned this pull request Jun 4, 2025

MQE: skip decoding histogram buckets where possible grafana/mimir#11587

Merged

1 task

charleskorn marked this pull request as ready for review June 4, 2025 04:33

charleskorn requested a review from roidelapluie as a code owner June 4, 2025 04:33

krajorama requested review from beorn7, bwplotka and krajorama June 4, 2025 08:05

beorn7 requested changes Jun 5, 2025

View reviewed changes

promql/histogram_stats_iterator.go Outdated Show resolved Hide resolved

promql/histogram_stats_iterator.go Show resolved Hide resolved

promql/histogram_stats_iterator.go Show resolved Hide resolved

charleskorn added 4 commits June 11, 2025 13:00

Merge branch 'main' into charleskorn/reuse-histogramstatsiterator

293c6da

# Conflicts: # promql/histogram_stats_iterator.go

Address PR feedback: add doc comment

55649f8

Signed-off-by: Charles Korn <charles.korn@grafana.com>

Address PR feedback: rename field

b88f19e

Signed-off-by: Charles Korn <charles.korn@grafana.com>

Address PR feedback: add doc comment for HistogramStatsIterator

1141530

Signed-off-by: Charles Korn <charles.korn@grafana.com>

charleskorn requested a review from beorn7 June 17, 2025 23:59

beorn7 approved these changes Jun 18, 2025

View reviewed changes

beorn7 merged commit 8d9dfa0 into prometheus:main Jun 18, 2025
27 checks passed

charleskorn mentioned this pull request Jun 19, 2025

[main] Merge upstream prometheus/main to 76a97ca273a7 grafana/mimir-prometheus#907

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

promql: reuse `histogramStatsIterator` where possible, and expose it for other implementations to use #16686

promql: reuse `histogramStatsIterator` where possible, and expose it for other implementations to use #16686

Uh oh!

charleskorn commented Jun 4, 2025 •

edited

Loading

Uh oh!

beorn7 commented Jun 4, 2025

Uh oh!

beorn7 left a comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

promql: reuse histogramStatsIterator where possible, and expose it for other implementations to use #16686

promql: reuse histogramStatsIterator where possible, and expose it for other implementations to use #16686

Uh oh!

Conversation

charleskorn commented Jun 4, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

beorn7 commented Jun 4, 2025

Uh oh!

beorn7 left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

promql: reuse `histogramStatsIterator` where possible, and expose it for other implementations to use #16686

promql: reuse `histogramStatsIterator` where possible, and expose it for other implementations to use #16686

charleskorn commented Jun 4, 2025 •

edited

Loading