Skip to content

Propagate consider_startup flag to compressed chunk table #9738

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
akuzm merged 6 commits into from
May 13, 2026
Merged

Propagate consider_startup flag to compressed chunk table #9738

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
3721a7b
Propagate consider_startup flag to compressed chunk table
akuzm May 7, 2026
463fe71
Merge remote-tracking branch 'origin/main' into HEAD
akuzm May 8, 2026
b2c0fb1
format
akuzm May 8, 2026
d0c2c31
support multicolumn distinct too
akuzm May 8, 2026
f73fa57
accept
akuzm May 11, 2026
a95cd29
Merge branch 'main' into consider-compressed
akuzm May 11, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
45 changes: 24 additions & 21 deletions tsl/src/nodes/columnar_scan/columnar_scan.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1175,6 +1175,27 @@ ts_columnar_scan_generate_paths(PlannerInfo *root, RelOptInfo *chunk_rel, const
chunk_rel->pathlist = NIL;
chunk_rel->partial_pathlist = NIL;

/*
* We want to consider startup costs so that IndexScan is preferred to
* sorted SeqScan when we may have a chance to use SkipScan. We consider
* startup costs for LIMIT queries, and SkipScan is basically a
* "LIMIT 1" query run "ndistinct" times. At this point we don't have
* all information to check if SkipScan can be used, but we can narrow
* it down.
*
* First, check if this query is candidate for SELECT DISTINCT SkipScan.
*/
const bool potential_select_distinct = list_length(root->distinct_pathkeys) >= 1;

/* Next, candidate for DISTINCT aggregate SkipScan */
const bool potential_distinct_aggregate =
root->numOrderedAggs >= 1 && list_length(root->group_pathkeys) == 1;

if (potential_select_distinct || potential_distinct_aggregate)
{
chunk_rel->consider_startup = true;
}

/* add RangeTblEntry and RelOptInfo for compressed chunk */
columnar_scan_add_plannerinfo(root,
compression_info,
Expand All @@ -1193,8 +1214,6 @@ ts_columnar_scan_generate_paths(PlannerInfo *root, RelOptInfo *chunk_rel, const

RelOptInfo *compressed_rel = compression_info->compressed_rel;

compressed_rel->consider_parallel = chunk_rel->consider_parallel;

/* translate chunk_rel->baserestrictinfo */
if (ts_guc_enable_columnar_scan_filter_pushdown)
{
Expand Down Expand Up @@ -1267,25 +1286,6 @@ ts_columnar_scan_generate_paths(PlannerInfo *root, RelOptInfo *chunk_rel, const
&sort_info,
compression_info);

/*
* We want to consider startup costs so that IndexScan is preferred to
* sorted SeqScan when we may have a chance to use SkipScan. We consider
* startup costs for LIMIT queries, and SkipScan is basically a
* "LIMIT 1" query run "ndistinct" times. At this point we don't have
* all information to check if SkipScan can be used, but we can narrow
* it down.
*/
if (!chunk_rel->consider_startup && IsA(compressed_path, IndexPath))
{
/* Candidate for SELECT DISTINCT SkipScan */
if (list_length(root->distinct_pathkeys) == 1
/* Candidate for DISTINCT aggregate SkipScan */
|| (root->numOrderedAggs >= 1 && list_length(root->group_pathkeys) == 1))
{
chunk_rel->consider_startup = true;
}
}

/*
* Add the paths to the chunk relation.
*/
Expand Down Expand Up @@ -2404,6 +2404,9 @@ columnar_scan_add_plannerinfo(PlannerInfo *root, CompressionInfo *info, const Ch
/* translate chunk_rel->joininfo for compressed_rel */
compressed_rel_setup_joininfo(compressed_rel, info);

compressed_rel->consider_parallel = chunk_rel->consider_parallel;
compressed_rel->consider_startup = chunk_rel->consider_startup;

/*
* Force parallel plan creation, see compute_parallel_worker().
* This is not compatible with ts_classify_relation(), but on the other hand
Expand Down
4 changes: 4 additions & 0 deletions tsl/test/expected/skip_scan.out
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2878,6 +2878,10 @@ psql:include/skip_scan_multi_query.sql:44: INFO: SkipScan used on compress_hype
psql:include/skip_scan_multi_query.sql:44: INFO: SkipScan used on compress_hyper_9_76_chunk_status_region_dev_dev_name__ts_me_idx(region NOT NULL, dev NOT NULL)
-- RowCompareExpr
:PREFIX SELECT DISTINCT ON (status, dev) * FROM :TABLE WHERE region = 'reg_1' and (status, dev) > (1,2);
psql:include/skip_scan_multi_query.sql:47: INFO: SkipScan used on compress_hyper_9_73_chunk_status_region_dev_dev_name__ts_me_idx(status NOT NULL, dev NOT NULL)
psql:include/skip_scan_multi_query.sql:47: INFO: SkipScan used on compress_hyper_9_74_chunk_status_region_dev_dev_name__ts_me_idx(status NOT NULL, dev NOT NULL)
psql:include/skip_scan_multi_query.sql:47: INFO: SkipScan used on compress_hyper_9_75_chunk_status_region_dev_dev_name__ts_me_idx(status NOT NULL, dev NOT NULL)
psql:include/skip_scan_multi_query.sql:47: INFO: SkipScan used on compress_hyper_9_76_chunk_status_region_dev_dev_name__ts_me_idx(status NOT NULL, dev NOT NULL)
-- always false expr similar to our initial skip qual
:PREFIX SELECT DISTINCT ON (status, region, dev) * FROM :TABLE WHERE dev > NULL and status > NULL and region > NULL;
-- no tuples matching
Expand Down
Loading