Repository navigation
Conversation
Assemble backend-local tuple and page measurements and take per-call snapshots of the existing index results. Access methods accumulate results across passes, so retain each pass's delta without counting earlier work again. Share the measurement helper between serial and parallel vacuum. Keep SP-GiST's newly deleted page count cumulative within one vacuum, without counting pages that were already empty on entry. Handle AMs that reset their newly deleted page count between passes. Adapt to PostgreSQL 16's existing recently_dead_tuples, missed_dead_tuples, pages_scanned, pages_removed and missed_dead_pages counters. Assemble these measurements independently of optional hooks; storage and SQL follow in separate reporting commits. Adapted-from: f2592a3
Introduce dead_pages in LVRelState and PgStat_VacuumStats. Count heap pages that retain recently dead tuples in both pruning and no-prune scans. Keep this measurement distinct from pages with dead tuples missed because of buffer pins. For index cleanup, derive the count from deleted pages that are not yet reusable. Cumulative statistics storage and SQL exposure follow separately. Adapted-from: 37a4753
Include pages_frozen and tuples_frozen in the backend-local VACUUM measurements. PostgreSQL 16 already counts frozen_pages and tuples_frozen and prints them in VACUUM VERBOSE, so reuse those counters without adding a second increment at the freezing site. Cumulative statistics storage and SQL exposure follow separately. Adapted-from: 93ae811
Add the backend-local pages_all_visible measurement. Count new visibility map marks under the heap page lock, including empty pages and second-pass cleanup. Test the current VM bit, rather than a potentially stale skipping decision, and do not recount existing marks or all-frozen-only upgrades. Cumulative statistics storage and SQL exposure follow separately. Adapted-from: 3aefc13
Remember the result of vacuum_get_cutoffs before DISABLE_PAGE_SKIPPING can force aggressive mode. Include this classification in backend-local work measurements, keeping freeze-age vacuums distinct from page-skipping options and wraparound failsafe activation. Cumulative statistics storage and SQL exposure follow separately. Adapted-from: cefc6ba
Count completed compactions, moved tuples, scanned pages and the final segment count without an extension hook. Discard state left over from an interrupted operation. Keep per-index work measurements for each pass. Native cumulative storage and SQL reporting follow separately. Adapted-from: 6f38fb2
Connect the previously introduced heap, index and AO measurements to ordinary relation and database statistics independently of any extension hook. Include them in native snapshots, resets, transactional drops and clean-shutdown persistence. Store dead_tuples and total_file_segs as last-run samples; other work counters accumulate. Exclude index reports and last-run samples from database totals. Advance the statistics-file format for the stored fields. Catalog functions, views and tests follow in a separate SQL reporting commit.
Publish cumulative tuple, page and AO compaction counters through native SQL getters and local, per-segment and cluster summary views. Keep last-run snapshots separate from cumulative work and normalize replicated relations. Include heap/index/AO tests for counters, resets, snapshots and persistence.
Reset work for one relation or the current database without clearing timing, invocation counts, other native counters, or extension metrics. Add a privileged gp_stat_reset_vacuum_stats wrapper for all primary segments and cover scope, permissions, pending work, and restart behavior.
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What does this PR do?
Make VACUUM work available through native cumulative statistics for tables,
indexes and databases.
Collect tuple removal, remaining and missed dead tuples, scanned and removed
pages, freezing work, newly all-visible pages, and freeze-age-driven vacuums.
For append-optimized tables, collect segment counts, completed compactions,
moved tuples and remaining hidden rows.
Report index work as per-call deltas and correct SP-GiST accounting for
previously empty pages. Database totals exclude separate index reports to
avoid double counting.
Store dead_tuples and total_file_segs as last-run samples; other work counters
accumulate. Use ordinary statistics snapshots, resets, DROP handling and
clean-shutdown persistence.
Measurement, cumulative storage, and catalog functions/views are introduced
in separate commits.
Impact
Makes VACUUM work statistics available for heap, AO/AOCO tables and indexes
without an extension. Native cumulative counters are updated when
track_counts is enabled and exposed through local and cluster-wide views.
Adds counter maintenance during VACUUM and increases relation and database
statistics entry sizes. Bumps the catalog and cumulative-statistics file
versions; development clusters initialized with the previous catalog version
require initdb.
Type of Change
Test Plan
make installcheckmake -C src/test installcheck-cbdb-parallelChecklist