diff --git a/doc/src/sgml/ref/alter_table.sgml b/doc/src/sgml/ref/alter_table.sgml index 8867da6c693..e0808afc350 100644 --- a/doc/src/sgml/ref/alter_table.sgml +++ b/doc/src/sgml/ref/alter_table.sgml @@ -376,22 +376,20 @@ WITH ( MODULUS numeric_literal, REM n_distinct_inherited, which override the number-of-distinct-values estimates made by subsequent ANALYZE - operations. n_distinct affects the statistics for the table - itself, while n_distinct_inherited affects the statistics - gathered for the table plus its inheritance children. When set to a - positive value, ANALYZE will assume that the column contains - exactly the specified number of distinct nonnull values. When set to a - negative value, which must be greater - than or equal to -1, ANALYZE will assume that the number of - distinct nonnull values in the column is linear in the size of the - table; the exact count is to be computed by multiplying the estimated - table size by the absolute value of the given number. For example, + operations. Ordinarily n_distinct is used. + n_distinct_inherited exists to allow the distinct + estimate to be overwritten for the statistics gathered for inheritance + parent tables and for partitioned tables. When the value specified is a + positive value, the query planner will assume that + the column contains exactly the specified number of distinct nonnull + values. Fractional values may also be specified by using values below 0 and above + or equal to -1. This instructs the query planner to estimate the number + of distinct values by multiplying the absolute value of the specified + number by the estimated number of rows in the table. For example, a value of -1 implies that all values in the column are distinct, while - a value of -0.5 implies that each value appears twice on the average. - This can be useful when the size of the table changes over time, since - the multiplication by the number of rows in the table is not performed - until query planning time. Specify a value of 0 to revert to estimating - the number of distinct values normally. For more information on the use + a value of -0.5 implies that each value appears twice on average. + This can be useful when the size of the table changes over time. + For more information on the use of statistics by the PostgreSQL query planner, refer to . diff --git a/src/backend/access/heap/heapam.c b/src/backend/access/heap/heapam.c index 568696333c2..c9f7d4aa1f7 100644 --- a/src/backend/access/heap/heapam.c +++ b/src/backend/access/heap/heapam.c @@ -490,6 +490,16 @@ heap_setscanlimits(TableScanDesc sscan, BlockNumber startBlk, BlockNumber numBlk scan->rs_startblock = startBlk; scan->rs_numblocks = numBlks; + + /* set the limits in the ParallelBlockTableScanDesc, when present as leader */ + if (scan->rs_base.rs_parallel != NULL && !IsParallelWorker()) + { + ParallelBlockTableScanDesc bpscan; + + bpscan = (ParallelBlockTableScanDesc) scan->rs_base.rs_parallel; + bpscan->phs_startblock = startBlk; + bpscan->phs_numblock = numBlks; + } } /* diff --git a/src/backend/access/table/tableam.c b/src/backend/access/table/tableam.c index 5e41404937e..5ad6963dbbe 100644 --- a/src/backend/access/table/tableam.c +++ b/src/backend/access/table/tableam.c @@ -188,6 +188,42 @@ table_beginscan_parallel(Relation relation, ParallelTableScanDesc pscan) pscan, flags); } +TableScanDesc +table_beginscan_parallel_tidrange(Relation relation, + ParallelTableScanDesc pscan, + ItemPointer mintid, ItemPointer maxtid) +{ + Snapshot snapshot; + uint32 flags = SO_TYPE_TIDRANGESCAN | SO_ALLOW_PAGEMODE; + TableScanDesc sscan; + + Assert(RelFileLocatorEquals(relation->rd_locator, pscan->phs_locator)); + + /* disable syncscan in parallel tid range scan. */ + pscan->phs_syncscan = false; + + if (!pscan->phs_snapshot_any) + { + /* Snapshot was serialized -- restore it */ + snapshot = RestoreSnapshot((char *) pscan + pscan->phs_snapshot_off); + RegisterSnapshot(snapshot); + flags |= SO_TEMP_SNAPSHOT; + } + else + { + /* SnapshotAny passed by caller (not serialized) */ + snapshot = SnapshotAny; + } + + sscan = relation->rd_tableam->scan_begin(relation, snapshot, 0, NULL, + pscan, flags); + + /* Set the scan limits based on the given min and max tids */ + relation->rd_tableam->scan_set_tidrange(sscan, mintid, maxtid); + + return sscan; +} + /* ---------------------------------------------------------------------------- * Index scan related functions. @@ -398,6 +434,7 @@ table_block_parallelscan_initialize(Relation rel, ParallelTableScanDesc pscan) bpscan->phs_nblocks > NBuffers / 4; SpinLockInit(&bpscan->phs_mutex); bpscan->phs_startblock = InvalidBlockNumber; + bpscan->phs_numblock = InvalidBlockNumber; pg_atomic_init_u64(&bpscan->phs_nallocated, 0); return sizeof(ParallelBlockTableScanDescData); @@ -577,8 +614,15 @@ table_block_parallelscan_nextpage(Relation rel, pbscanwork->phsw_chunk_remaining = pbscanwork->phsw_chunk_size - 1; } + /* + * Check if we've allocated every block in the relation, or if we've + * reached the limit imposed by pbscan->phs_numblock (if set). + */ if (nallocated >= pbscan->phs_nblocks) - page = InvalidBlockNumber; /* all blocks have been allocated */ + page = InvalidBlockNumber; /* all blocks have been allocated */ + else if (pbscan->phs_numblock != InvalidBlockNumber && + nallocated >= pbscan->phs_numblock) + page = InvalidBlockNumber; /* upper scan limit reached */ else page = (nallocated + pbscan->phs_startblock) % pbscan->phs_nblocks; diff --git a/src/backend/executor/execParallel.c b/src/backend/executor/execParallel.c index f098a5557cf..c8c41cac1ea 100644 --- a/src/backend/executor/execParallel.c +++ b/src/backend/executor/execParallel.c @@ -40,6 +40,7 @@ #include "executor/nodeSeqscan.h" #include "executor/nodeSort.h" #include "executor/nodeSubplan.h" +#include "executor/nodeTidrangescan.h" #include "executor/tqueue.h" #include "jit/jit.h" #include "nodes/nodeFuncs.h" @@ -266,6 +267,11 @@ ExecParallelEstimate(PlanState *planstate, ExecParallelEstimateContext *e) ExecForeignScanEstimate((ForeignScanState *) planstate, e->pcxt); break; + case T_TidRangeScanState: + if (planstate->plan->parallel_aware) + ExecTidRangeScanEstimate((TidRangeScanState *) planstate, + e->pcxt); + break; case T_AppendState: if (planstate->plan->parallel_aware) ExecAppendEstimate((AppendState *) planstate, @@ -493,6 +499,11 @@ ExecParallelInitializeDSM(PlanState *planstate, ExecForeignScanInitializeDSM((ForeignScanState *) planstate, d->pcxt); break; + case T_TidRangeScanState: + if (planstate->plan->parallel_aware) + ExecTidRangeScanInitializeDSM((TidRangeScanState *) planstate, + d->pcxt); + break; case T_AppendState: if (planstate->plan->parallel_aware) ExecAppendInitializeDSM((AppendState *) planstate, @@ -994,6 +1005,11 @@ ExecParallelReInitializeDSM(PlanState *planstate, ExecForeignScanReInitializeDSM((ForeignScanState *) planstate, pcxt); break; + case T_TidRangeScanState: + if (planstate->plan->parallel_aware) + ExecTidRangeScanReInitializeDSM((TidRangeScanState *) planstate, + pcxt); + break; case T_AppendState: if (planstate->plan->parallel_aware) ExecAppendReInitializeDSM((AppendState *) planstate, pcxt); @@ -1362,6 +1378,11 @@ ExecParallelInitializeWorker(PlanState *planstate, ParallelWorkerContext *pwcxt) ExecForeignScanInitializeWorker((ForeignScanState *) planstate, pwcxt); break; + case T_TidRangeScanState: + if (planstate->plan->parallel_aware) + ExecTidRangeScanInitializeWorker((TidRangeScanState *) planstate, + pwcxt); + break; case T_AppendState: if (planstate->plan->parallel_aware) ExecAppendInitializeWorker((AppendState *) planstate, pwcxt); diff --git a/src/backend/executor/nodeTidrangescan.c b/src/backend/executor/nodeTidrangescan.c index 1bce8d6cbfe..065bd37d9dd 100644 --- a/src/backend/executor/nodeTidrangescan.c +++ b/src/backend/executor/nodeTidrangescan.c @@ -128,14 +128,17 @@ TidExprListCreate(TidRangeScanState *tidrangestate) * TidRangeEval * * Compute and set node's block and offset range to scan by evaluating - * node->trss_tidexprs. Returns false if we detect the range cannot - * contain any tuples. Returns true if it's possible for the range to - * contain tuples. We don't bother validating that trss_mintid is less - * than or equal to trss_maxtid, as the scan_set_tidrange() table AM - * function will handle that. + * node->trss_tidexprs. Sets node's trss_rangeIsEmpty to true if the + * calculated range must be empty of any tuples, otherwise sets + * trss_rangeIsEmpty to false and sets trss_mintid and trss_maxtid to the + * calculated range. + * + * We don't bother validating that trss_mintid is less than or equal to + * trss_maxtid, as the scan_set_tidrange() table AM function will handle + * that. * ---------------------------------------------------------------- */ -static bool +static void TidRangeEval(TidRangeScanState *node) { ExprContext *econtext = node->ss.ps.ps_ExprContext; @@ -165,7 +168,10 @@ TidRangeEval(TidRangeScanState *node) /* If the bound is NULL, *nothing* matches the qual. */ if (isNull) - return false; + { + node->trss_rangeIsEmpty = true; + return; + } if (tidopexpr->exprtype == TIDEXPR_LOWER_BOUND) { @@ -207,7 +213,7 @@ TidRangeEval(TidRangeScanState *node) ItemPointerCopy(&lowerBound, &node->trss_mintid); ItemPointerCopy(&upperBound, &node->trss_maxtid); - return true; + node->trss_rangeIsEmpty = false; } /* ---------------------------------------------------------------- @@ -234,12 +240,19 @@ TidRangeNext(TidRangeScanState *node) slot = node->ss.ss_ScanTupleSlot; direction = estate->es_direction; - if (!node->trss_inScan) + /* First time through, compute TID range to scan */ + if (!node->trss_rangeCalcDone) { - /* First time through, compute TID range to scan */ - if (!TidRangeEval(node)) - return NULL; + TidRangeEval(node); + node->trss_rangeCalcDone = true; + } + /* Check if the range was detected not to contain any tuples */ + if (node->trss_rangeIsEmpty) + return NULL; + + if (!node->trss_inScan) + { if (scandesc == NULL) { scandesc = table_beginscan_tidrange(node->ss.ss_currentRelation, @@ -250,9 +263,13 @@ TidRangeNext(TidRangeScanState *node) } else { - /* rescan with the updated TID range */ - table_rescan_tidrange(scandesc, &node->trss_mintid, - &node->trss_maxtid); + /* rescan with the updated TID range only in non-parallel mode */ + if (scandesc->rs_parallel == NULL) + { + /* rescan with the updated TID range */ + table_rescan_tidrange(scandesc, &node->trss_mintid, + &node->trss_maxtid); + } } node->trss_inScan = true; @@ -274,13 +291,18 @@ TidRangeNext(TidRangeScanState *node) static bool TidRangeRecheck(TidRangeScanState *node, TupleTableSlot *slot) { - if (!TidRangeEval(node)) - return false; + /* First call? Compute the TID Range */ + if (!node->trss_rangeCalcDone) + { + TidRangeEval(node); + node->trss_rangeCalcDone = true; + } Assert(ItemPointerIsValid(&slot->tts_tid)); /* Recheck the ctid is still within range */ - if (ItemPointerCompare(&slot->tts_tid, &node->trss_mintid) < 0 || + if (node->trss_rangeIsEmpty || + ItemPointerCompare(&slot->tts_tid, &node->trss_mintid) < 0 || ItemPointerCompare(&slot->tts_tid, &node->trss_maxtid) > 0) return false; @@ -322,6 +344,9 @@ ExecReScanTidRangeScan(TidRangeScanState *node) /* mark scan as not in progress, and tid range list as not computed yet */ node->trss_inScan = false; + /* mark that the TID range needs to be recalculated */ + node->trss_rangeCalcDone = false; + /* * We must wait until TidRangeNext before calling table_rescan_tidrange. */ @@ -380,6 +405,10 @@ ExecInitTidRangeScan(TidRangeScan *node, EState *estate, int eflags) * mark scan as not in progress, and TID range as not computed yet */ tidrangestate->trss_inScan = false; + tidrangestate->trss_rangeCalcDone = false; + + /* This will be calculated correctly in TidRangeEval() */ + tidrangestate->trss_rangeIsEmpty = true; /* * open the scan relation @@ -415,3 +444,109 @@ ExecInitTidRangeScan(TidRangeScan *node, EState *estate, int eflags) */ return tidrangestate; } +/* ---------------------------------------------------------------- + * Parallel Scan Support + * ---------------------------------------------------------------- + */ + +/* ---------------------------------------------------------------- + * ExecTidRangeScanEstimate + * + * Compute the amount of space we'll need in the parallel + * query DSM, and inform pcxt->estimator about our needs. + * ---------------------------------------------------------------- + */ +void +ExecTidRangeScanEstimate(TidRangeScanState *node, ParallelContext *pcxt) +{ + EState *estate = node->ss.ps.state; + + node->trss_pscanlen = + table_parallelscan_estimate(node->ss.ss_currentRelation, + estate->es_snapshot); + shm_toc_estimate_chunk(&pcxt->estimator, node->trss_pscanlen); + shm_toc_estimate_keys(&pcxt->estimator, 1); +} + +/* ---------------------------------------------------------------- + * ExecTidRangeScanInitializeDSM + * + * Set up a parallel TID scan descriptor. + * ---------------------------------------------------------------- + */ +void +ExecTidRangeScanInitializeDSM(TidRangeScanState *node, ParallelContext *pcxt) +{ + EState *estate = node->ss.ps.state; + ParallelTableScanDesc pscan; + + pscan = shm_toc_allocate(pcxt->toc, node->trss_pscanlen); + table_parallelscan_initialize(node->ss.ss_currentRelation, + pscan, + estate->es_snapshot); + shm_toc_insert(pcxt->toc, node->ss.ps.plan->plan_node_id, pscan); + + /* Determine the TID Range */ + TidRangeEval(node); + node->trss_rangeCalcDone = true; + + /* Setup the scandesc, unless TidRangeEval found the range is empty */ + if (!node->trss_rangeIsEmpty) + node->ss.ss_currentScanDesc = + table_beginscan_parallel_tidrange(node->ss.ss_currentRelation, + pscan, + &node->trss_mintid, + &node->trss_maxtid); + else + node->ss.ss_currentScanDesc = NULL; +} + +/* ---------------------------------------------------------------- + * ExecTidRangeScanReInitializeDSM + * + * Reset shared state before beginning a fresh scan. + * ---------------------------------------------------------------- + */ +void +ExecTidRangeScanReInitializeDSM(TidRangeScanState *node, + ParallelContext *pcxt) +{ + ParallelTableScanDesc pscan; + + pscan = node->ss.ss_currentScanDesc->rs_parallel; + table_parallelscan_reinitialize(node->ss.ss_currentRelation, pscan); + + /* Determine the TID Range */ + TidRangeEval(node); + node->trss_rangeCalcDone = true; + + if (!node->trss_rangeIsEmpty) + node->ss.ss_currentRelation->rd_tableam->scan_set_tidrange(node->ss.ss_currentScanDesc, + &node->trss_mintid, + &node->trss_maxtid); +} + +/* ---------------------------------------------------------------- + * ExecTidRangeScanInitializeWorker + * + * Copy relevant information from TOC into planstate. + * ---------------------------------------------------------------- + */ +void +ExecTidRangeScanInitializeWorker(TidRangeScanState *node, + ParallelWorkerContext *pwcxt) +{ + ParallelTableScanDesc pscan; + + pscan = shm_toc_lookup(pwcxt->toc, node->ss.ps.plan->plan_node_id, false); + + /* Determine the TID Range */ + TidRangeEval(node); + node->trss_rangeCalcDone = true; + + if (!node->trss_rangeIsEmpty) + node->ss.ss_currentScanDesc = table_beginscan_parallel_tidrange(node->ss.ss_currentRelation, + pscan, + &node->trss_mintid, + &node->trss_maxtid); +} diff --git a/src/backend/optimizer/path/costsize.c b/src/backend/optimizer/path/costsize.c index 94077e6a006..03ac556c73d 100644 --- a/src/backend/optimizer/path/costsize.c +++ b/src/backend/optimizer/path/costsize.c @@ -1366,8 +1366,9 @@ cost_tidrangescan(Path *path, PlannerInfo *root, { Selectivity selectivity; double pages; - Cost startup_cost = 0; - Cost run_cost = 0; + Cost startup_cost; + Cost cpu_run_cost; + Cost disk_run_cost; QualCost qpqual_cost; Cost cpu_per_tuple; QualCost tid_qual_cost; @@ -1396,11 +1397,7 @@ cost_tidrangescan(Path *path, PlannerInfo *root, /* * The first page in a range requires a random seek, but each subsequent - * page is just a normal sequential page read. NOTE: it's desirable for - * TID Range Scans to cost more than the equivalent Sequential Scans, - * because Seq Scans have some performance advantages such as scan - * synchronization and parallelizability, and we'd prefer one of them to - * be picked unless a TID Range Scan really is better. + * page is just a normal sequential page read. */ ntuples = selectivity * baserel->tuples; nseqpages = pages - 1.0; @@ -1417,7 +1414,7 @@ cost_tidrangescan(Path *path, PlannerInfo *root, &spc_seq_page_cost); /* disk costs; 1 random page and the remainder as seq pages */ - run_cost += spc_random_page_cost + spc_seq_page_cost * nseqpages; + disk_run_cost = spc_random_page_cost + spc_seq_page_cost * nseqpages; /* Add scanning CPU costs */ get_restriction_qual_cost(root, baserel, param_info, &qpqual_cost); @@ -1429,20 +1426,35 @@ cost_tidrangescan(Path *path, PlannerInfo *root, * can't be removed, this is a mistake and we're going to underestimate * the CPU cost a bit.) */ - startup_cost += qpqual_cost.startup + tid_qual_cost.per_tuple; + startup_cost = qpqual_cost.startup + tid_qual_cost.per_tuple; cpu_per_tuple = cpu_tuple_cost + qpqual_cost.per_tuple - tid_qual_cost.per_tuple; - run_cost += cpu_per_tuple * ntuples; + cpu_run_cost = cpu_per_tuple * ntuples; /* tlist eval costs are paid per output row, not per tuple scanned */ startup_cost += path->pathtarget->cost.startup; - run_cost += path->pathtarget->cost.per_tuple * path->rows; + cpu_run_cost += path->pathtarget->cost.per_tuple * path->rows; + + /* Adjust costing for parallelism, if used. */ + if (path->parallel_workers > 0) + { + double parallel_divisor = get_parallel_divisor(path); + + /* The CPU cost is divided among all the workers. */ + cpu_run_cost /= parallel_divisor; + + /* + * In the case of a parallel plan, the row count needs to represent + * the number of tuples processed per worker. + */ + path->rows = clamp_row_est(path->rows / parallel_divisor); + } /* we should not generate this path type when enable_tidscan=false */ Assert(enable_tidscan); path->disabled_nodes = 0; path->startup_cost = startup_cost; - path->total_cost = startup_cost + run_cost; + path->total_cost = startup_cost + cpu_run_cost + disk_run_cost; } /* diff --git a/src/backend/optimizer/path/tidpath.c b/src/backend/optimizer/path/tidpath.c index 2bfb338b81c..e9e52e4abe7 100644 --- a/src/backend/optimizer/path/tidpath.c +++ b/src/backend/optimizer/path/tidpath.c @@ -492,7 +492,8 @@ ec_member_matches_ctid(PlannerInfo *root, RelOptInfo *rel, * create_tidscan_paths * Create paths corresponding to direct TID scans of the given rel. * - * Candidate paths are added to the rel's pathlist (using add_path). + * Candidate paths are added to the rel's pathlist (using add_path) and to + * partial_pathlist via add_partial_path, in the case of partial paths. */ bool create_tidscan_paths(PlannerInfo *root, RelOptInfo *rel) @@ -553,7 +554,24 @@ create_tidscan_paths(PlannerInfo *root, RelOptInfo *rel) add_path(rel, (Path *) create_tidrangescan_path(root, rel, tidrangequals, - required_outer)); + required_outer, + 0)); + + /* If appropriate, consider parallel tid range scan. */ + if (rel->consider_parallel && required_outer == NULL) + { + int parallel_workers; + + parallel_workers = compute_parallel_worker(rel, rel->pages, -1, + max_parallel_workers_per_gather); + + if (parallel_workers > 0) + add_partial_path(rel, (Path *) create_tidrangescan_path(root, + rel, + tidrangequals, + required_outer, + parallel_workers)); + } } /* diff --git a/src/backend/optimizer/plan/planner.c b/src/backend/optimizer/plan/planner.c index e8ea78c0c97..4f2a7ddd23f 100644 --- a/src/backend/optimizer/plan/planner.c +++ b/src/backend/optimizer/plan/planner.c @@ -1173,12 +1173,14 @@ subquery_planner(PlannerGlobal *glob, Query *parse, char *plan_name, foreach(l, (List *) parse->havingQual) { Node *havingclause = (Node *) lfirst(l); + Relids having_relids; if (contain_agg_clause(havingclause) || contain_volatile_functions(havingclause) || contain_subplans(havingclause) || (parse->groupClause && parse->groupingSets && - bms_is_member(root->group_rtindex, pull_varnos(root, havingclause)))) + ((having_relids = pull_varnos(root, havingclause)) == NULL || + bms_is_member(root->group_rtindex, having_relids)))) { /* keep it in HAVING */ newHaving = lappend(newHaving, havingclause); diff --git a/src/backend/optimizer/util/pathnode.c b/src/backend/optimizer/util/pathnode.c index bca51b4067b..1bb613d44b0 100644 --- a/src/backend/optimizer/util/pathnode.c +++ b/src/backend/optimizer/util/pathnode.c @@ -1261,7 +1261,8 @@ create_tidscan_path(PlannerInfo *root, RelOptInfo *rel, List *tidquals, */ TidRangePath * create_tidrangescan_path(PlannerInfo *root, RelOptInfo *rel, - List *tidrangequals, Relids required_outer) + List *tidrangequals, Relids required_outer, + int parallel_workers) { TidRangePath *pathnode = makeNode(TidRangePath); @@ -1270,9 +1271,9 @@ create_tidrangescan_path(PlannerInfo *root, RelOptInfo *rel, pathnode->path.pathtarget = rel->reltarget; pathnode->path.param_info = get_baserel_parampathinfo(root, rel, required_outer); - pathnode->path.parallel_aware = false; + pathnode->path.parallel_aware = (parallel_workers > 0); pathnode->path.parallel_safe = rel->consider_parallel; - pathnode->path.parallel_workers = 0; + pathnode->path.parallel_workers = parallel_workers; pathnode->path.pathkeys = NIL; /* always unordered */ pathnode->tidrangequals = tidrangequals; diff --git a/src/include/access/relscan.h b/src/include/access/relscan.h index b5e0fb386c0..3da43557a13 100644 --- a/src/include/access/relscan.h +++ b/src/include/access/relscan.h @@ -96,6 +96,8 @@ typedef struct ParallelBlockTableScanDescData BlockNumber phs_nblocks; /* # blocks in relation at start of scan */ slock_t phs_mutex; /* mutual exclusion for setting startblock */ BlockNumber phs_startblock; /* starting block number */ + BlockNumber phs_numblock; /* # blocks to scan, or InvalidBlockNumber if + * no limit */ pg_atomic_uint64 phs_nallocated; /* number of blocks allocated to * workers so far. */ } ParallelBlockTableScanDescData; diff --git a/src/include/access/tableam.h b/src/include/access/tableam.h index e16bf025692..3c241d10a52 100644 --- a/src/include/access/tableam.h +++ b/src/include/access/tableam.h @@ -1130,6 +1130,18 @@ extern void table_parallelscan_initialize(Relation rel, extern TableScanDesc table_beginscan_parallel(Relation relation, ParallelTableScanDesc pscan); +/* + * Begin a parallel tidrange scan. `pscan` needs to have been initialized with + * table_parallelscan_initialize(), for the same relation. The initialization + * does not need to have happened in this backend. + * + * Caller must hold a suitable lock on the relation. + */ +extern TableScanDesc table_beginscan_parallel_tidrange(Relation relation, + ParallelTableScanDesc pscan, + ItemPointer mintid, + ItemPointer maxtid); + /* * Restart a parallel scan. Call this in the leader process. Caller is * responsible for making sure that all workers have finished the scan diff --git a/src/include/executor/nodeTidrangescan.h b/src/include/executor/nodeTidrangescan.h index a831f1202ca..2b5465b3ce4 100644 --- a/src/include/executor/nodeTidrangescan.h +++ b/src/include/executor/nodeTidrangescan.h @@ -14,6 +14,7 @@ #ifndef NODETIDRANGESCAN_H #define NODETIDRANGESCAN_H +#include "access/parallel.h" #include "nodes/execnodes.h" extern TidRangeScanState *ExecInitTidRangeScan(TidRangeScan *node, @@ -21,4 +22,10 @@ extern TidRangeScanState *ExecInitTidRangeScan(TidRangeScan *node, extern void ExecEndTidRangeScan(TidRangeScanState *node); extern void ExecReScanTidRangeScan(TidRangeScanState *node); +/* parallel scan support */ +extern void ExecTidRangeScanEstimate(TidRangeScanState *node, ParallelContext *pcxt); +extern void ExecTidRangeScanInitializeDSM(TidRangeScanState *node, ParallelContext *pcxt); +extern void ExecTidRangeScanReInitializeDSM(TidRangeScanState *node, ParallelContext *pcxt); +extern void ExecTidRangeScanInitializeWorker(TidRangeScanState *node, ParallelWorkerContext *pwcxt); + #endif /* NODETIDRANGESCAN_H */ diff --git a/src/include/nodes/execnodes.h b/src/include/nodes/execnodes.h index a36653c37f9..c6e3ccff046 100644 --- a/src/include/nodes/execnodes.h +++ b/src/include/nodes/execnodes.h @@ -1924,6 +1924,9 @@ typedef struct TidScanState * trss_mintid the lowest TID in the scan range * trss_maxtid the highest TID in the scan range * trss_inScan is a scan currently in progress? + * trss_rangeCalcDone has the TID range been calculated yet? + * trss_rangeIsEmpty true if the TID range is certainly empty + * trss_pscanlen size of parallel heap scan descriptor * ---------------- */ typedef struct TidRangeScanState @@ -1933,6 +1936,9 @@ typedef struct TidRangeScanState ItemPointerData trss_mintid; ItemPointerData trss_maxtid; bool trss_inScan; + bool trss_rangeCalcDone; + bool trss_rangeIsEmpty; + Size trss_pscanlen; } TidRangeScanState; /* ---------------- diff --git a/src/include/optimizer/pathnode.h b/src/include/optimizer/pathnode.h index da60383c2aa..090e836cfc9 100644 --- a/src/include/optimizer/pathnode.h +++ b/src/include/optimizer/pathnode.h @@ -67,7 +67,8 @@ extern TidPath *create_tidscan_path(PlannerInfo *root, RelOptInfo *rel, extern TidRangePath *create_tidrangescan_path(PlannerInfo *root, RelOptInfo *rel, List *tidrangequals, - Relids required_outer); + Relids required_outer, + int parallel_workers); extern AppendPath *create_append_path(PlannerInfo *root, RelOptInfo *rel, List *subpaths, List *partial_subpaths, List *pathkeys, Relids required_outer, diff --git a/src/test/regress/expected/tidrangescan.out b/src/test/regress/expected/tidrangescan.out index 721f3b94e04..bbb1b87fa89 100644 --- a/src/test/regress/expected/tidrangescan.out +++ b/src/test/regress/expected/tidrangescan.out @@ -297,4 +297,110 @@ FETCH LAST c; COMMIT; DROP TABLE tidrangescan; +-- Tests for parallel tidrangescans +SET parallel_setup_cost TO 0; +SET parallel_tuple_cost TO 0; +SET min_parallel_table_scan_size TO 0; +SET max_parallel_workers_per_gather TO 4; +CREATE TABLE parallel_tidrangescan(id integer, data text) WITH (fillfactor = 10); +-- Insert enough tuples such that each page gets 5 tuples with fillfactor = 10 +INSERT INTO parallel_tidrangescan SELECT i, repeat('x', 100) FROM generate_series(1,200) AS s(i); +-- Ensure there are 40 pages for parallel test +SELECT min(ctid), max(ctid) FROM parallel_tidrangescan; + min | max +-------+-------- + (0,1) | (39,5) +(1 row) + +-- Parallel range scans with upper bound +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid < '(30,1)'; + QUERY PLAN +-------------------------------------------------------------------- + Finalize Aggregate + -> Gather + Workers Planned: 4 + -> Partial Aggregate + -> Parallel Tid Range Scan on parallel_tidrangescan + TID Cond: (ctid < '(30,1)'::tid) +(6 rows) + +SELECT count(*) FROM parallel_tidrangescan WHERE ctid < '(30,1)'; + count +------- + 150 +(1 row) + +-- Parallel range scans with lower bound +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)'; + QUERY PLAN +-------------------------------------------------------------------- + Finalize Aggregate + -> Gather + Workers Planned: 4 + -> Partial Aggregate + -> Parallel Tid Range Scan on parallel_tidrangescan + TID Cond: (ctid > '(10,0)'::tid) +(6 rows) + +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)'; + count +------- + 150 +(1 row) + +-- Parallel range scans with both bounds +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)' AND ctid < '(30,1)'; + QUERY PLAN +----------------------------------------------------------------------------------- + Finalize Aggregate + -> Gather + Workers Planned: 4 + -> Partial Aggregate + -> Parallel Tid Range Scan on parallel_tidrangescan + TID Cond: ((ctid > '(10,0)'::tid) AND (ctid < '(30,1)'::tid)) +(6 rows) + +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)' AND ctid < '(30,1)'; + count +------- + 100 +(1 row) + +-- Parallel rescans +EXPLAIN (COSTS OFF) +SELECT t.ctid,t2.c FROM parallel_tidrangescan t, +LATERAL (SELECT count(*) c FROM parallel_tidrangescan t2 WHERE t2.ctid <= t.ctid) t2 +WHERE t.ctid < '(1,0)'; + QUERY PLAN +---------------------------------------------------------------- + Nested Loop + -> Gather + Workers Planned: 4 + -> Parallel Tid Range Scan on parallel_tidrangescan t + TID Cond: (ctid < '(1,0)'::tid) + -> Aggregate + -> Tid Range Scan on parallel_tidrangescan t2 + TID Cond: (ctid <= t.ctid) +(8 rows) + +SELECT t.ctid,t2.c FROM parallel_tidrangescan t, +LATERAL (SELECT count(*) c FROM parallel_tidrangescan t2 WHERE t2.ctid <= t.ctid) t2 +WHERE t.ctid < '(1,0)'; + ctid | c +-------+--- + (0,1) | 1 + (0,2) | 2 + (0,3) | 3 + (0,4) | 4 + (0,5) | 5 +(5 rows) + +DROP TABLE parallel_tidrangescan; +RESET max_parallel_workers_per_gather; +RESET min_parallel_table_scan_size; +RESET parallel_tuple_cost; +RESET parallel_setup_cost; RESET enable_seqscan; diff --git a/src/test/regress/sql/tidrangescan.sql b/src/test/regress/sql/tidrangescan.sql index ac09ebb6262..f86493530e2 100644 --- a/src/test/regress/sql/tidrangescan.sql +++ b/src/test/regress/sql/tidrangescan.sql @@ -98,4 +98,49 @@ COMMIT; DROP TABLE tidrangescan; +-- Tests for parallel tidrangescans +SET parallel_setup_cost TO 0; +SET parallel_tuple_cost TO 0; +SET min_parallel_table_scan_size TO 0; +SET max_parallel_workers_per_gather TO 4; + +CREATE TABLE parallel_tidrangescan(id integer, data text) WITH (fillfactor = 10); + +-- Insert enough tuples such that each page gets 5 tuples with fillfactor = 10 +INSERT INTO parallel_tidrangescan SELECT i, repeat('x', 100) FROM generate_series(1,200) AS s(i); + +-- Ensure there are 40 pages for parallel test +SELECT min(ctid), max(ctid) FROM parallel_tidrangescan; + +-- Parallel range scans with upper bound +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid < '(30,1)'; +SELECT count(*) FROM parallel_tidrangescan WHERE ctid < '(30,1)'; + +-- Parallel range scans with lower bound +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)'; +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)'; + +-- Parallel range scans with both bounds +EXPLAIN (COSTS OFF) +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)' AND ctid < '(30,1)'; +SELECT count(*) FROM parallel_tidrangescan WHERE ctid > '(10,0)' AND ctid < '(30,1)'; + +-- Parallel rescans +EXPLAIN (COSTS OFF) +SELECT t.ctid,t2.c FROM parallel_tidrangescan t, +LATERAL (SELECT count(*) c FROM parallel_tidrangescan t2 WHERE t2.ctid <= t.ctid) t2 +WHERE t.ctid < '(1,0)'; + +SELECT t.ctid,t2.c FROM parallel_tidrangescan t, +LATERAL (SELECT count(*) c FROM parallel_tidrangescan t2 WHERE t2.ctid <= t.ctid) t2 +WHERE t.ctid < '(1,0)'; + +DROP TABLE parallel_tidrangescan; + +RESET max_parallel_workers_per_gather; +RESET min_parallel_table_scan_size; +RESET parallel_tuple_cost; +RESET parallel_setup_cost; RESET enable_seqscan;