v41-0003-PS-ExprState-cache-modifications.patch
application/octet-stream
Filename: v41-0003-PS-ExprState-cache-modifications.patch
Type: application/octet-stream
Part: 0
Patch
Same data as JSON:
GET /api/v1/attachments/:id/patch
the parsed metadata as JSON — format, series position, per-file stats; never the diff bytes.
API reference →
Format: format-patch
Series: patch v41-0003
Subject: PS - ExprState cache modifications.
| File | + | − |
|---|---|---|
| src/backend/replication/pgoutput/pgoutput.c | 139 | 75 |
From 7c8e8a9c89b9fb7c10dc3a3ec560e3c21c0c1a8f Mon Sep 17 00:00:00 2001
From: Ajin Cherian <ajinc@fast.au.fujitsu.com>
Date: Mon, 22 Nov 2021 20:26:50 -0500
Subject: [PATCH v41 3/6] PS - ExprState cache modifications.
Now the cached row-filters (e.g. ExprState *) are invalidated only in
rel_sync_cache_relation_cb function, so it means the ALTER PUBLICATION for one
table should not cause row-filters of other tables to also become invalidated.
Also all code related to caching row-filters has been removed from the
get_rel_sync_entry function and is now done just before they are needed in the
pgoutput_row_filter function.
If there are multiple publication filters for a given table these are are all
combined into a single filter.
Author: Peter Smith, Greg Nancarrow
Changes are based on a suggestions from Amit [1] [2], and Houz [3]
[1] https://www.postgresql.org/message-id/CAA4eK1%2BxQb06NGs6Y7OzwMtKYYixEqR8tdWV5THAVE4SAqNrDg%40mail.gmail.com
[2] https://www.postgresql.org/message-id/CAA4eK1%2Btio46goUKBUfAKFsFVxtgk8nOty%3DTxKoKH-gdLzHD2g%40mail.gmail.com
[3] https://www.postgresql.org/message-id/OS0PR01MB5716090A70A73ADF58C58950948D9%40OS0PR01MB5716.jpnprd01.prod.outlook.com
---
src/backend/replication/pgoutput/pgoutput.c | 214 ++++++++++++++++++----------
1 file changed, 139 insertions(+), 75 deletions(-)
diff --git a/src/backend/replication/pgoutput/pgoutput.c b/src/backend/replication/pgoutput/pgoutput.c
index 3643684..fd024d4 100644
--- a/src/backend/replication/pgoutput/pgoutput.c
+++ b/src/backend/replication/pgoutput/pgoutput.c
@@ -1,4 +1,4 @@
-/*-------------------------------------------------------------------------
+/*------------------------------------------------------------------------
*
* pgoutput.c
* Logical Replication output plugin
@@ -21,6 +21,7 @@
#include "executor/executor.h"
#include "fmgr.h"
#include "nodes/nodeFuncs.h"
+#include "nodes/makefuncs.h"
#include "optimizer/optimizer.h"
#include "parser/parse_coerce.h"
#include "replication/logical.h"
@@ -123,7 +124,16 @@ typedef struct RelationSyncEntry
bool replicate_valid;
PublicationActions pubactions;
- List *exprstate; /* ExprState for row filter */
+
+ /*
+ * Row-filter related members:
+ * The flag 'rowfilter_valid' indicates if the exprstate has been assigned
+ * yet or not. We cannot just use the exprstate value for this purpose
+ * because there might be no filter at all for the current relid (e.g.
+ * exprstate is NULL).
+ */
+ bool rowfilter_valid;
+ ExprState *exprstate; /* ExprState for row filter(s) */
TupleTableSlot *scantuple; /* tuple table slot for row filter */
/*
@@ -161,7 +171,7 @@ static bool get_schema_sent_in_streamed_txn(RelationSyncEntry *entry,
static EState *create_estate_for_relation(Relation rel);
static ExprState *pgoutput_row_filter_init_expr(Node *rfnode);
static bool pgoutput_row_filter_exec_expr(ExprState *state, ExprContext *econtext);
-static bool pgoutput_row_filter(Relation relation, HeapTuple oldtuple,
+static bool pgoutput_row_filter(PGOutputData *data, Relation relation, HeapTuple oldtuple,
HeapTuple newtuple, RelationSyncEntry *entry);
/*
@@ -731,20 +741,118 @@ pgoutput_row_filter_exec_expr(ExprState *state, ExprContext *econtext)
* If it returns true, the change is replicated, otherwise, it is not.
*/
static bool
-pgoutput_row_filter(Relation relation, HeapTuple oldtuple, HeapTuple newtuple, RelationSyncEntry *entry)
+pgoutput_row_filter(PGOutputData *data, Relation relation, HeapTuple oldtuple, HeapTuple newtuple, RelationSyncEntry *entry)
{
EState *estate;
ExprContext *ecxt;
ListCell *lc;
bool result = true;
+ Oid relid = RelationGetRelid(relation);
+ List *rfnodes = NIL;
+ int n_filters;
+
+ /*
+ * If the row filter caching is currently flagged "invalid" then it means we
+ * don't know yet if there is/isn't any row filters for this relation.
+ *
+ * This code is usually one-time execution.
+ *
+ * NOTE: The ExprState cache could have been created up-front in the
+ * function get_rel_sync_entry() instead of the deferred on-the-fly
+ * assignment below. The reason for choosing to do it here is because there
+ * are some scenarios where the get_rel_sync_entry() is called but where a
+ * row will not be published. For example, for truncate, we may not need
+ * any row evaluation, so there is no need to compute it. It would also be
+ * a waste if any error happens before actually evaluating the filter. And
+ * tomorrow there could be other operations (which use get_rel_sync_entry)
+ * but which don't need to build ExprState. So the decision was to defer
+ * this logic to last moment when we know it will be needed.
+ */
+ if (!entry->rowfilter_valid)
+ {
+ MemoryContext oldctx;
+ TupleDesc tupdesc = RelationGetDescr(relation);
+
+ /*
+ * Create a tuple table slot for row filter. TupleDesc must live as
+ * long as the cache remains. Release the tuple table slot if it
+ * already exists.
+ */
+ if (entry->scantuple != NULL)
+ {
+ ExecDropSingleTupleTableSlot(entry->scantuple);
+ entry->scantuple = NULL;
+ }
+ oldctx = MemoryContextSwitchTo(CacheMemoryContext);
+ tupdesc = CreateTupleDescCopy(tupdesc);
+ entry->scantuple = MakeSingleTupleTableSlot(tupdesc, &TTSOpsHeapTuple);
+ MemoryContextSwitchTo(oldctx);
+
+ /*
+ * Find if there are any row filters for this relation. If there are,
+ * then prepare the necessary ExprState and cache it in entry->exprstate.
+ *
+ * NOTE: All publication-table mappings must be checked.
+ *
+ * NOTE: If the relation is a partition and pubviaroot is true, use
+ * the row filter of the topmost partitioned table instead of the row
+ * filter of its own partition.
+ */
+ foreach(lc, data->publications)
+ {
+ Publication *pub = lfirst(lc);
+ HeapTuple rftuple;
+ Datum rfdatum;
+ bool rfisnull;
+
+ /*
+ * Lookup if there is a row-filter, and if yes remember it in a list.
+ * In code following this 'publications' loop we will combine all filters.
+ */
+ rftuple = SearchSysCache2(PUBLICATIONRELMAP, ObjectIdGetDatum(entry->publish_as_relid), ObjectIdGetDatum(pub->oid));
+ if (HeapTupleIsValid(rftuple))
+ {
+ rfdatum = SysCacheGetAttr(PUBLICATIONRELMAP, rftuple, Anum_pg_publication_rel_prqual, &rfisnull);
+
+ if (!rfisnull)
+ {
+ Node *rfnode;
+
+ oldctx = MemoryContextSwitchTo(CacheMemoryContext);
+ rfnode = stringToNode(TextDatumGetCString(rfdatum));
+ rfnodes = lappend(rfnodes, rfnode);
+ MemoryContextSwitchTo(oldctx);
+ }
+
+ ReleaseSysCache(rftuple);
+ }
+
+ } /* loop all subscribed publications */
+
+ /*
+ * Combine all the row-filters (if any) into a single filter, and then build the ExprState for it
+ */
+ n_filters = list_length(rfnodes);
+ if (n_filters > 0)
+ {
+ Node *rfnode;
+
+ oldctx = MemoryContextSwitchTo(CacheMemoryContext);
+ rfnode = n_filters > 1 ? makeBoolExpr(AND_EXPR, rfnodes, -1) : linitial(rfnodes);
+ entry->exprstate = pgoutput_row_filter_init_expr(rfnode);
+ MemoryContextSwitchTo(oldctx);
+ }
+
+ entry->rowfilter_valid = true;
+ }
/* Bail out if there is no row filter */
- if (entry->exprstate == NIL)
+ if (!entry->exprstate)
return true;
elog(DEBUG3, "table \"%s.%s\" has row filter",
- get_namespace_name(get_rel_namespace(RelationGetRelid(relation))),
- get_rel_name(relation->rd_id));
+ get_namespace_name(get_rel_namespace(relid)),
+ get_rel_name(relid));
PushActiveSnapshot(GetTransactionSnapshot());
@@ -757,20 +865,13 @@ pgoutput_row_filter(Relation relation, HeapTuple oldtuple, HeapTuple newtuple, R
ExecStoreHeapTuple(newtuple ? newtuple : oldtuple, ecxt->ecxt_scantuple, false);
/*
- * If the subscription has multiple publications and the same table has a
- * different row filter in these publications, all row filters must be
- * matched in order to replicate this change.
+ * NOTE: Multiple publication row-filters have already been combined to a
+ * single exprstate.
*/
- foreach(lc, entry->exprstate)
+ if (entry->exprstate)
{
- ExprState *exprstate = (ExprState *) lfirst(lc);
-
/* Evaluates row filter */
- result = pgoutput_row_filter_exec_expr(exprstate, ecxt);
-
- /* If the tuple does not match one of the row filters, bail out */
- if (!result)
- break;
+ result = pgoutput_row_filter_exec_expr(entry->exprstate, ecxt);
}
/* Cleanup allocated resources */
@@ -840,7 +941,7 @@ pgoutput_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
HeapTuple tuple = &change->data.tp.newtuple->tuple;
/* Check row filter. */
- if (!pgoutput_row_filter(relation, NULL, tuple, relentry))
+ if (!pgoutput_row_filter(data, relation, NULL, tuple, relentry))
break;
/*
@@ -873,7 +974,7 @@ pgoutput_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
HeapTuple newtuple = &change->data.tp.newtuple->tuple;
/* Check row filter. */
- if (!pgoutput_row_filter(relation, oldtuple, newtuple, relentry))
+ if (!pgoutput_row_filter(data, relation, oldtuple, newtuple, relentry))
break;
maybe_send_schema(ctx, change, relation, relentry);
@@ -907,7 +1008,7 @@ pgoutput_change(LogicalDecodingContext *ctx, ReorderBufferTXN *txn,
HeapTuple oldtuple = &change->data.tp.oldtuple->tuple;
/* Check row filter. */
- if (!pgoutput_row_filter(relation, oldtuple, NULL, relentry))
+ if (!pgoutput_row_filter(data, relation, oldtuple, NULL, relentry))
break;
maybe_send_schema(ctx, change, relation, relentry);
@@ -1321,10 +1422,11 @@ get_rel_sync_entry(PGOutputData *data, Relation relation)
entry->schema_sent = false;
entry->streamed_txns = NIL;
entry->replicate_valid = false;
+ entry->rowfilter_valid = false;
entry->pubactions.pubinsert = entry->pubactions.pubupdate =
entry->pubactions.pubdelete = entry->pubactions.pubtruncate = false;
entry->scantuple = NULL;
- entry->exprstate = NIL;
+ entry->exprstate = NULL;
entry->publish_as_relid = InvalidOid;
entry->map = NULL; /* will be set by maybe_send_schema() if
* needed */
@@ -1344,7 +1446,6 @@ get_rel_sync_entry(PGOutputData *data, Relation relation)
List *schemaPubids = GetSchemaPublications(schemaId);
ListCell *lc;
Oid publish_as_relid = relid;
- TupleDesc tupdesc = RelationGetDescr(relation);
/* Reload publications if needed before use. */
if (!publications_valid)
@@ -1358,22 +1459,6 @@ get_rel_sync_entry(PGOutputData *data, Relation relation)
publications_valid = true;
}
- /* Release tuple table slot */
- if (entry->scantuple != NULL)
- {
- ExecDropSingleTupleTableSlot(entry->scantuple);
- entry->scantuple = NULL;
- }
-
- /*
- * Create a tuple table slot for row filter. TupleDesc must live as
- * long as the cache remains.
- */
- oldctx = MemoryContextSwitchTo(CacheMemoryContext);
- tupdesc = CreateTupleDescCopy(tupdesc);
- entry->scantuple = MakeSingleTupleTableSlot(tupdesc, &TTSOpsHeapTuple);
- MemoryContextSwitchTo(oldctx);
-
/*
* Build publication cache. We can't use one provided by relcache as
* relcache considers all publications given relation is in, but here
@@ -1383,9 +1468,6 @@ get_rel_sync_entry(PGOutputData *data, Relation relation)
{
Publication *pub = lfirst(lc);
bool publish = false;
- HeapTuple rftuple;
- Datum rfdatum;
- bool rfisnull;
if (pub->alltables)
{
@@ -1449,33 +1531,6 @@ get_rel_sync_entry(PGOutputData *data, Relation relation)
entry->pubactions.pubtruncate |= pub->pubactions.pubtruncate;
}
- /*
- * Cache row filter, if available. All publication-table mappings
- * must be checked. If it is a partition and pubviaroot is true,
- * use the row filter of the topmost partitioned table instead of
- * the row filter of its own partition.
- */
- rftuple = SearchSysCache2(PUBLICATIONRELMAP, ObjectIdGetDatum(publish_as_relid), ObjectIdGetDatum(pub->oid));
- if (HeapTupleIsValid(rftuple))
- {
- rfdatum = SysCacheGetAttr(PUBLICATIONRELMAP, rftuple, Anum_pg_publication_rel_prqual, &rfisnull);
-
- if (!rfisnull)
- {
- Node *rfnode;
- ExprState *exprstate;
-
- oldctx = MemoryContextSwitchTo(CacheMemoryContext);
- rfnode = stringToNode(TextDatumGetCString(rfdatum));
-
- /* Prepare for expression execution */
- exprstate = pgoutput_row_filter_init_expr(rfnode);
- entry->exprstate = lappend(entry->exprstate, exprstate);
- MemoryContextSwitchTo(oldctx);
- }
-
- ReleaseSysCache(rftuple);
- }
}
list_free(pubids);
@@ -1582,6 +1637,21 @@ rel_sync_cache_relation_cb(Datum arg, Oid relid)
free_conversion_map(entry->map);
}
entry->map = NULL;
+
+ /*
+ * Row filter cache cleanups. (Will be rebuilt later if needed).
+ */
+ entry->rowfilter_valid = false;
+ if (entry->scantuple != NULL)
+ {
+ ExecDropSingleTupleTableSlot(entry->scantuple);
+ entry->scantuple = NULL;
+ }
+ if (entry->exprstate != NULL)
+ {
+ pfree(entry->exprstate);
+ entry->exprstate = NULL;
+ }
}
}
@@ -1622,12 +1692,6 @@ rel_sync_cache_publication_cb(Datum arg, int cacheid, uint32 hashvalue)
entry->pubactions.pubupdate = false;
entry->pubactions.pubdelete = false;
entry->pubactions.pubtruncate = false;
-
- if (entry->exprstate != NIL)
- {
- list_free_deep(entry->exprstate);
- entry->exprstate = NIL;
- }
}
MemoryContextSwitchTo(oldctx);
--
1.8.3.1