Release v2.29.2

• fix(pgfr_record): tolerate missing UPDATE privilege on cron.job (#69) (ea72b3c)
• feat(site): one-click install page on GitHub Pages (#68) (6f4a3aa)

Release v2.29.1

Highlights

📋 New single-file install bundle for the Supabase SQL editor. Two new release assets — pgfr_record-v2.29.1-bundle.sql and pgfr_analyze-v2.29.1-bundle.sql — are self-contained, copy-pasteable scripts with zero psql metacommands. Open the Supabase dashboard SQL editor, paste, run. No psql, no CLI, no extension registry needed. (#66)

🔍 XID, MultiXID, and xmin horizon monitoring. First-class wraparound observability for both transaction-ID counters and configurable warning/critical thresholds, plus point-in-time identification of the backend holding the cluster's xmin horizon. (#53, #54)

What's new

Features

• Single-file install bundle for SQL-editor channels (#66)
• xmin horizon monitoring (#53)
• MultiXID wraparound monitoring + configurable XID/MXID thresholds (#54)

Refactors

• Consolidate all pg_cron scheduling into pgfr_record.enable() (#57, #58) — one canonical entry point, easier to reason about during incidents
• Standardize pg_cron job naming on the pgfr_ underscore prefix (#59, #62)

Fixes

• Warn when cron.log_run is on, since pgfr's job cadence makes cron.job_run_details grow unbounded (#63)
• Drain query_map before the "empty initially" assertion to remove a pg_cron race in test_ring_buffer.sql (#65)

Docs

• README + explainer page consistency pass — split recorder self-check from database diagnostics, accurate Architecture flow, no more bogus "superuser required" requirement (#64)
• Apache 2.0 license + SPDX headers + license-aware bundling

Installation

# psql users
psql --single-transaction -f pgfr_record/install.sql
psql --single-transaction -f pgfr_analyze/install.sql

# Supabase dashboard / any SQL editor: download the *-bundle.sql assets
#   from this release and paste them in.

# dbdev / CREATE EXTENSION users
select dbdev.install('dventimi-pgfr_record');
select dbdev.install('dventimi-pgfr_analyze');
create extension pgfr_record;
create extension pgfr_analyze;

About v2.29.1

v2.29.1 is a build-pipeline retag of v2.29 with no code differences. v2.29's dbdev publish was rejected by the dbdev API for declaring requires = 'pgfr_record' in pgfr_analyze's control file. v2.29.1 strips that line inside the dbdev publish path only — the source control file keeps the declaration so direct CREATE EXTENSION pgfr_analyze users still get a clean dependency error if pgfr_record is absent. (#67)

Release v2.29

⚠️ Superseded by v2.29.1. v2.29.1 is a build-pipeline retag with no code differences; v2.29's dbdev publish failed. See v2.29.1 for the readable release announcement and a working dbdev publish.

---

• feat(build): add single-file install bundle for SQL-editor channel (#66) (0247725)
• fix(test): drain query_map before "empty initially" assertion (#65) (5e42651)
• docs: README + explainer page consistency pass (#64) (de00b20)
• (feat) add xmin horizon monitoring (#53) (f741d6f)
• Warn when cron.log_run leaves job_run_details unbounded (#63) (52448f5)
• chore: add SPDX headers, license-aware bundling, CONTRIBUTING (d6b9eca)
• chore: add Apache 2.0 license (7f3ac50)
• docs: align hero terminal mock with real health_check() output (73b56af)
• docs: remove remaining "high-frequency" claims from landing page (b4e3654)
• docs: correct architecture blurb on landing page (045430e)
• docs: refresh landing page stats to reflect current code (84f3811)
• docs: rename pg-flight-recorder -> pg_flight_recorder (578f348)
• Minor formatting improvements in PG_COMPAT.md (9fd8986)
• feat(pgfr): monitor MultiXID wraparound + configurable XID/MXID thresholds (#54) (c79456c)
• refactor(pgfr_record): standardize pg_cron job names on underscore prefix (#59) (b128a7f)
• refactor(pgfr_record): consolidate all pg_cron scheduling into enable() (#57) (3d0bdc2)
• fix(ci): deactivate pgfr_ cron jobs immediately after install (race with #46) (733ee67)
• fix(ci): tighten readiness probe with SELECT 1 instead of pg_isready (6b4197e)
• fix(pgfr_analyze): declare hard dependency on pgfr_record in control file (bf8a17c)
• docs(site): fix inaccurate claims in index.html explainer (ea48806)
• Minor formatting of tables (8dc0437)
• Minor formatting of tables (903439e)
• Minor formatting of tables (b6ac744)
• Killed the blog (85278f9)
• fix(pgfr_record): add statement_timeout to all maintenance cron jobs (#49) (1769333)
• docs: fix inaccurate claims in README.md and REFERENCE.md (2e52b1b)
• docs(REFERENCE): fix retention profile row — wrong key name and wrong troubleshooting value (fed4316)
• fix(ci): pin pg_cron/pgtap and make CI builds verbose (#44) (93c6ed5)
• fix(ci): make PostgreSQL 18 test matrix actually pass (f6d80ba)
• fix(pgfr_record): add DEFAULT partition to all range-partitioned tables (25481e2)
• fix(ci): clear lint errors on main (6aa78cb)
• fix(ci): inline \ir directives when publishing to dbdev (f96003f)
• refactor: switch install.sql to \ir relative include (e10058e)
• feat(ci): add PostgreSQL 18 to the test matrix (03b0670)
• docs: add blog post on stacked PRs in the AI age (cd2203d)
• refactor: move experimental pgfr_control to feature branch (7af4f46)
• refactor: rename _record/, _analyze/, _control/ to pgfr_record/, pgfr_analyze/, pgfr_control/ (17451ab)
• fix(record): lock collection works with snapshot_based = false (f7bd95b)
• fix(analyze): report() backend_type → backend_state for v2 wait_summary (852dda6)
• fix(analyze): use delta columns in storm/regression detection (d5e932c)
• fix(ring): truncate activity_samples on rotation (d9889fb)
• fix: PG13/PG15/PG18 compat, docker-compose, test fixes (c8f5832)
• chore: update install.sql entry points to source modular SQL files (a3818bf)
• refactor(_analyze) + feat: split into modules, add v2 reader functions (6a348d6)
• feat(phase3): daily-partitioned snapshot tables + migration script (889b76a)
• feat(phase2): N-partition TRUNCATE ring buffer (wait/activity/lock/query samples) (f948d80)
• feat(phase1): sparse collectors, partition infra, migration path (7970fde)
• refactor(_record): split monolithic install.sql into focused modules (8300152)
• chore: benchmark scripts, PG18 compat notes, Dockerfile (26c6dca)
• docs(blueprints): storage overhaul spec — zero-bloat partitioned design (bac1144)
• Let quickstart steps span full container width (bd9d650)
• Stack quickstart steps vertically to prevent horizontal overflow (65cbb7e)
• Use

   instead of  for quickstart blocks to properly display multi-line content (5bd6401)

• Add copy buttons to quickstart code blocks and clean up formatting (22c49d7)
• Add link to project website in README (4d1b297)
• Add GitHub Pages deployment workflow (67cde00)
• Add landing page with Supabase brand identity (c089234)
• Killed some pointless comments (2b97742)

Release v2.28.1

• mothball benchmark and .claude directories (9e77e65)
• fix REFERENCE.md schema version and statements_interval defaults (2865b4e)
• add missing COMMENT ON statements for 4 analyze functions (91e50c6)
• add missing COMMENT ON statements for 14 public functions (3b2a1ae)
• fix ring retention to use configured slots instead of hardcoded 120 (5544b0e)
• fix profile table defaults, enable() fallback, and stale comments (58166dc)
• fix remaining stale interval references across codebase (165b9bf)
• update stale interval/retention values to match 60s default (1f5da31)
• fix stale install notice and adjust retention warning thresholds (b57f222)
• change default snapshot and statement intervals to every minute (a121a53)
• change default sample interval from 3 minutes to 1 minute (423dc52)
• rename core schema from pgfr to pgfr_record for consistency (fde159e)
• move analysis functions from pgfr_record internals to pgfr_analyze (7fcef53)
• update docs to reflect renamed functions and enforce package boundaries (5511e5b)
• update _record/README.md to reference pgfr_analyze wrapper names (9537366)
• internalize analysis functions in pgfr_record, add user-facing API in pgfr_analyze (d607a2b)
• add delta columns and improve pg_stat_statements collection (544b66e)
• remove SQL_INDEX.json and its supporting infrastructure (20cb531)
• update REFERENCE.md: fix stale skip reason and add missing safety features (63e36fb)
• update top-level README safety table to reflect current features (f3bca6d)
• add outer statement_timeout to all pg_cron jobs (89def08)
• regenerate SQL_INDEX.json (7f994d4)
• remove backup detection from pre-flight checks (32705ca)
• Minor formatting changes (6513664)
• regenerate SQL_INDEX.json (7ea2a07)
• fix backup detection self-matching in _should_skip_collection() (4f4ad57)
• regenerate SQL_INDEX.json (7e1fe0c)
• remove broken/dead safety features: DDL lock check, load throttle, walsender false-positive, collection jitter, job deduplication, adaptive sampling, and auto-mode adjustment (45b926c)
• regenerate SQL_INDEX.json (76f299c)
• remove broken checkpoint detection from pre-flight checks (b558f47)
• regenerate SQL_INDEX.json (1e9ca2a)
• remove dead replica lag check from pre-flight safety gates (651e991)

Release v2.28

• drop individual file assets from release (archives are sufficient) (803426f)
• include _control in release assets and remove redundant flat copies (975b3d2)
• add supabase/ to .gitignore (3e61c5c)
• rewrite REFERENCE.md as exhaustive lookup reference (2fa2a3d)
• rewrite top-level README as project landing page (bd0b51a)
• make sub-package READMEs self-contained for dbdev publishing (8de16b1)
• use fully-qualified URL for top-level README link in _record/README.md (291cbc6)

Release v2.27

• regenerate SQL_INDEX.json (4337b7c)
• remove obsolete table of contents from _record/install.sql (2341df8)
• normalize markdown table formatting in README.md and REFERENCE.md (bcc55ec)
• update sql-index workflow to commit SQL_INDEX.json to main (ff2922f)
• track SQL_INDEX.json in main branch, regenerate for current structure (084d0a4)
• remove vestigial migrations directory (7119c24)
• update CLAUDE.md with per-extension compose file structure (0fcbca6)
• update REFERENCE.md for three-extension structure (f891456)
• decompose docker-compose.yml into per-extension compose files (2276250)
• update README.md with project structure table (3a464ac)
• update CLAUDE.md to reflect new project structure (01354f3)
• move control functions to pgfr_control schema, simplify uninstalls (4708c54)
• distribute tests among extension subdirectories (4a2d157)
• move uninstall files into subdirs, rename .control to extension.control (ba5dadf)
• move SQL files into their extension subdirectories (146ec5b)
• rename pg_flight_recorder → pgfr across all extensions and schemas (f647461)

Release v2.25

• remove blog/, tests/PATHOLOGY_TESTS.md, requirements.txt (51c17cb)
• Added the Resent deck (8bbaa82)
• remove pglite directory, was severely out of date (5f6d6c0)
• remove migrations directory, simplify upgrade path (565d861)
• remove tools/sql-find, query SQL_INDEX.json directly (bf2b9eb)
• fix sql-index to recover from per-statement parse errors (1bf4d65)
• bump schema_version to 2.25, add migration for reporting schema move (35ce340)
• move reporting functions to flight_recorder_reporting schema (df514b5)
• docs: clarify two install methods in README (3369e8d)
• move badges from README.md to REFERENCE.md for database.dev compatibility (06952fe)
• rename pg-flight-recorder to pg_flight_recorder for extension naming consistency (a0e7327)
• docs: update REFERENCE.md for core/reporting split (7002a35)
• docs: update install instructions for core/reporting split (3a4781b)
• feat: split reporting functions into separate file and two dbdev packages (423b264)
• fix(ci): use correct credentials.toml format for dbdev (9d92fc9)
• ci: extract publish-dbdev into standalone workflow (0ded618)
• fix(ci): use GitHub API to find dbdev CLI download URL (7691f48)

Release v2.19

• fix: remove psql-specific \set command for Supabase compatibility (5e37114)

v2.18 - Extension-Free Bloat Tracking

What's New

Features

• Extension-free table bloat tracking - Track table bloat without requiring pgstattuple extension
• Flight recorder skill - Added /flight-recorder skill for database analysis

Improvements

• Major codebase simplification - Removed ~7000 lines of code while preserving functionality
• Config consolidation - Eliminated config duplication with single source of truth
• Documentation rewrite - Completely rewritten README.md and REFERENCE.md for clarity

Fixes

• Fixed test reliability by disabling adaptive_sampling during tests

Export Enhancements

• Added snapshot_deltas example for SQLite delta computation
• Added _schema table to exports to prevent AI query errors

Release v2.17

• docs: remove DIAGNOSTIC_PLAYBOOKS.md (0f0799c)
• docs: restore DIAGNOSTIC_PLAYBOOKS.md (47fb089)
• chore: remove outdated PATHOLOGY_WORK_IN_PROGRESS.md (400d3a0)
• chore: remove outdated documentation and archive directory (8249c08)
• feat: replace timing-based metrics with buffer-based metrics as default (1b420c4)
• docs: add comprehensive in-database documentation for AI analysis (cbdd1b1)
• feat: add SQLite export function for AI-driven analysis (a1de5c1)
• docs: add section on hierarchical presentation to balance "record everything" (1fc6e3f)
• docs: expand profile documentation with detailed settings tables (c06d1df)
• feat: expand configuration profiles with 77 of 112 parameters (85bf40d)
• docs: document all 112 configuration parameters (23727ce)
• feat: add new feature parameters to configuration profiles (bcf1710)
• docs: add Canary Queries documentation to REFERENCE.md (cd43346)
• fix: correct markdown link fragments in archived feature ideas (dbdf9e5)
• docs: archive completed feature ideas, add future directions (3563b36)
• ci: add version badge and automated release workflow (864fbbe)
• docs: update documentation for complete 9/9 pathology coverage (88da7c0)
• fix: correct column name in anomaly_report query (details -> description) (92dc5f5)
• feat: complete all 9 pathology generators (48 tests) (66017b5)
• docs: update documentation for 6 pathologies (33 tests) (cc0b24b)
• fix: make pathology tests more lenient for CI environment (ff0f821)
• feat: add Connection Exhaustion pathology using dblink + fix SQL error (c96f97d)
• feat: add 3 more pathology generators (High CPU, Slow Realtime, Slow Queries) (60d6333)
• docs: fix markdown lint errors in documentation files (94b7738)
• docs: add session memory for pathology generator work (f6b28ea)
• docs: add PR description for pathology generators (4802ad6)
• feat: add pathological data generators for diagnostic validation (3ecafb5)