What metrico/gigapipe shipped
Written by FoxPlug from public releases; not affiliated with Gigapipe. An automatic summary of the public release, pull request and commit data of github.com/metrico/gigapipe. Gigapipe did not write it and does not use or endorse FoxPlug. Every line links to the public change it describes.
Get a weekly update like this for your product, free
Week of September 21, 2026
What shipped
- v5.5.2 released with fixes to promql range functions, logql grouping-less aggregations, and documentation updates. Release
- Fixed logql grouping-less aggregations that returned empty result sets instead of computed values since v5.4.3. Pull request #1000
- Gigapipe now accepts GIGAPIPE_-prefixed environment variables for configuration while maintaining backward compatibility with QRYN_ and CLOKI_ names. Pull request #994
- Fixed promql range functions to evaluate over the correct window (t-range, t] instead of a shifted interval. Pull request #984
- Fixed promql key selector bucket keying to use ceiling values consistent with other read paths. Pull request #998
- Clarified bucket sizing for range functions to be based on range rather than step. Pull request #983
- Pinned grpc dependency to v1.84.0 release to avoid pre-release snapshots. Pull request #1002
- Updated branding references to use Gigapipe name in documentation and project assets. Pull request #993
- Updated e2e test suite references to point to renamed gigapipe-tests repository. Pull request #992
- Changed ci runner from self-hosted to ubuntu-latest. Pull request #1001
Why it matters
This release resolves critical bugs in promql range function evaluation and logql aggregation results that affected query correctness. Configuration now standardizes on GIGAPIPE_ environment variables while keeping existing deployments working, and the project branding has been unified across documentation.
Changelog entry
- v5.5.2 Release
- fix(logql): return the empty label set for a grouping-less aggregation Pull request #1000
- feat(config): accept GIGAPIPE_ environment variables Pull request #994
- fix(promql): evaluate range functions over (t-range, t], not a shifted window Pull request #984
- fix(promql): key selector buckets by the ceiling, like every other read Pull request #998
- fix(promql): one definition of the range-function bucket cap Pull request #983
- deps: pin grpc to the v1.84.0 release Pull request #1002
- docs: add the compression dimension to
samples_v3sort key guidance Pull request #985 - chore: use the Gigapipe name where the old names are branding Pull request #993
- chore(e2e): point at gigapipe-tests and de-duplicate its path Pull request #992
v5.5.2 ships fixes for promql range functions and logql grouping-less aggregations. Configuration now uses GIGAPIPE_ environment variables. Backward compatible with existing deployments.
Gigapipe v5.5.2 is out. This release fixes critical issues with promql range function evaluation and logql aggregations that were returning incorrect empty results. The project now standardizes on GIGAPIPE_-prefixed environment variables for configuration—existing deployments continue working with the old QRYN_ and CLOKI_ prefixes. Documentation and project branding have been updated to reflect Gigapipe across the board.
Week of September 14, 2026
What shipped
- v5.5.1 released with fixes for PromQL range functions returning empty results and
query_rangeend parameter rounding. Release - Fixed rate(), irate(), increase(), delta(), deriv(), and related functions that returned no series when step >= range [1]. Pull request #981
- Fixed
query_rangeend parameter to floor to the 15s grid instead of ceiling, preventing over-extended query ranges [2]. Pull request #979 - Documented ClickHouse table ordering and ADVANCED_SAMPLES_ORDERING configuration with defaults and behavior [3]. Pull request #982
- v5.5.0 released with compression codecs for samples, traces, profiles tables and optional
metrics_15saggregation. Release - Added METRICS_15S_ENABLED flag to opt metrics out of the
metrics_15saggregate table, reducing disk cost at 15s scrape resolution [7]. Pull request #978 - Added compression codecs to samples, traces, and profiles tables, achieving 37% disk reduction on
samples_v3[8]. Pull request #977 - Fixed Elastic ingest to properly index _index and _id labels instead of storing empty values [6]. Pull request #975
- SmartBuffer now grows its RAM chunk on demand instead of preallocating 5MB for all reader endpoints [5]. Pull request #960
Why it matters
These releases fix critical issues with PromQL range functions and query parameter handling that affected correctness of results. Storage efficiency improvements through compression and optional aggregation tables help reduce infrastructure costs. The fixes make the system more reliable for both metric and log workloads.
Changelog entry
- v5.5.1: fix PromQL range functions,
query_rangeend rounding, document ClickHouse ordering Release - rate/irate/increase/delta/deriv/resets no longer return empty results when step >= range Pull request #981
query_rangeend parameter now floors to 15s grid instead of ceiling Pull request #979- Documented ADVANCED_SAMPLES_ORDERING and ClickHouse table ordering behavior Pull request #982
- v5.5.0: add compression codecs to samples, traces, profiles, patterns tables Release
- Add METRICS_15S_ENABLED flag to optionally disable
metrics_15saggregate table Pull request #978 - Compression codecs reduce
samples_v3size by 37%,tempo_traces_attrs_ginby 34% Pull request #977 - Elastic ingest now properly populates _index and _id labels Pull request #975
- SmartBuffer grows RAM chunk on demand instead of preallocating 5MB Pull request #960
Gigapipe v5.5.1: fixed PromQL range functions returning empty results, corrected query_range end rounding, and documented ClickHouse table ordering. v5.5.0 added compression codecs (37% smaller samples) and optional metrics_15s aggregation.
Gigapipe v5.5.1 and v5.5.0 ship critical fixes and improvements: PromQL range functions now return correct results, query_range end parameter rounds correctly, and compression codecs reduce disk usage by 37% on samples tables. New METRICS_15S_ENABLED flag lets operators opt out of expensive aggregation at 15s scrape resolution. Elastic ingest now properly indexes _index and _id labels.