Add the "snapshot too old" feature

This feature is controlled by a new old_snapshot_threshold GUC.  A
value of -1 disables the feature, and that is the default.  The
value of 0 is just intended for testing.  Above that it is the
number of minutes a snapshot can reach before pruning and vacuum
are allowed to remove dead tuples which the snapshot would
otherwise protect.  The xmin associated with a transaction ID does
still protect dead tuples.  A connection which is using an "old"
snapshot does not get an error unless it accesses a page modified
recently enough that it might not be able to produce accurate
results.

This is similar to the Oracle feature, and we use the same SQLSTATE
and error message for compatibility.

Author: kgrittn

contrib/bloom/blscan.c

@@ -138,7 +138,8 @@ blgetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
 									blkno, RBM_NORMAL, bas);
 
 		LockBuffer(buffer, BUFFER_LOCK_SHARE);
-		page = BufferGetPage(buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(buffer, scan->xs_snapshot, scan->indexRelation,
+							 BGP_TEST_FOR_OLD_SNAPSHOT);
 
 		if (!BloomPageIsDeleted(page))
 		{

doc/src/sgml/config.sgml

@@ -2041,6 +2041,42 @@ include_dir 'conf.d'
         </para>
        </listitem>
       </varlistentry>
+
+      <varlistentry id="guc-old-snapshot-threshold" xreflabel="old_snapshot_threshold">
+       <term><varname>old_snapshot_threshold</varname> (<type>integer</type>)
+       <indexterm>
+        <primary><varname>old_snapshot_threshold</> configuration parameter</primary>
+       </indexterm>
+       </term>
+       <listitem>
+        <para>
+         Sets the minimum time that a snapshot can be used without risk of a
+         <literal>snapshot too old</> error occurring when using the snapshot.
+         This parameter can only be set at server start.
+        </para>
+
+        <para>
+         Beyond the threshold, old data may be vacuumed away.  This can help
+         prevent bloat in the face of snapshots which remain in use for a
+         long time.  To prevent incorrect results due to cleanup of data which
+         would otherwise be visible to the snapshot, an error is generated
+         when the snapshot is older than this threshold and the snapshot is
+         used to read a page which has been modified since the snapshot was
+         built.
+        </para>
+
+        <para>
+         A value of <literal>-1</> disables this feature, and is the default.
+         Useful values for production work probably range from a small number
+         of hours to a few days.  The setting will be coerced to a granularity
+         of minutes, and small numbers (such as <literal>0</> or
+         <literal>1min</>) are only allowed because they may sometimes be
+         useful for testing.  While a setting as high as <literal>60d</> is
+         allowed, please note that in many workloads extreme bloat or
+         transaction ID wraparound may occur in much shorter time frames.
+        </para>
+       </listitem>
+      </varlistentry>
      </variablelist>
     </sect2>
    </sect1>
@@ -3051,6 +3087,10 @@ include_dir 'conf.d'
         You should also consider setting <varname>hot_standby_feedback</>
         on standby server(s) as an alternative to using this parameter.
        </para>
+       <para>
+        This does not prevent cleanup of dead rows which have reached the age
+        specified by <varname>old_snapshot_threshold</>.
+       </para>
       </listitem>
      </varlistentry>
 
@@ -3198,6 +3238,16 @@ include_dir 'conf.d'
         until it eventually reaches the primary.  Standbys make no other use
         of feedback they receive other than to pass upstream.
        </para>
+       <para>
+        This setting does not override the behavior of
+        <varname>old_snapshot_threshold</> on the primary; a snapshot on the
+        standby which exceeds the primary's age threshold can become invalid,
+        resulting in cancellation of transactions on the standby.  This is
+        because <varname>old_snapshot_threshold</> is intended to provide an
+        absolute limit on the time which dead rows can contribute to bloat,
+        which would otherwise be violated because of the configuration of a
+        standby.
+       </para>
       </listitem>
      </varlistentry>
 

src/backend/access/brin/brin.c

@@ -135,7 +135,7 @@ brininsert(Relation idxRel, Datum *values, bool *nulls,
 	MemoryContext tupcxt = NULL;
 	MemoryContext oldcxt = NULL;
 
-	revmap = brinRevmapInitialize(idxRel, &pagesPerRange);
+	revmap = brinRevmapInitialize(idxRel, &pagesPerRange, NULL);
 
 	for (;;)
 	{
@@ -152,7 +152,7 @@ brininsert(Relation idxRel, Datum *values, bool *nulls,
 		/* normalize the block number to be the first block in the range */
 		heapBlk = (heapBlk / pagesPerRange) * pagesPerRange;
 		brtup = brinGetTupleForHeapBlock(revmap, heapBlk, &buf, &off, NULL,
-										 BUFFER_LOCK_SHARE);
+										 BUFFER_LOCK_SHARE, NULL);
 
 		/* if range is unsummarized, there's nothing to do */
 		if (!brtup)
@@ -285,7 +285,8 @@ brinbeginscan(Relation r, int nkeys, int norderbys)
 	scan = RelationGetIndexScan(r, nkeys, norderbys);
 
 	opaque = (BrinOpaque *) palloc(sizeof(BrinOpaque));
-	opaque->bo_rmAccess = brinRevmapInitialize(r, &opaque->bo_pagesPerRange);
+	opaque->bo_rmAccess = brinRevmapInitialize(r, &opaque->bo_pagesPerRange,
+											   scan->xs_snapshot);
 	opaque->bo_bdesc = brin_build_desc(r);
 	scan->opaque = opaque;
 
@@ -368,7 +369,8 @@ bringetbitmap(IndexScanDesc scan, TIDBitmap *tbm)
 		MemoryContextResetAndDeleteChildren(perRangeCxt);
 
 		tup = brinGetTupleForHeapBlock(opaque->bo_rmAccess, heapBlk, &buf,
-									   &off, &size, BUFFER_LOCK_SHARE);
+									   &off, &size, BUFFER_LOCK_SHARE,
+									   scan->xs_snapshot);
 		if (tup)
 		{
 			tup = brin_copy_tuple(tup, size);
@@ -647,7 +649,7 @@ brinbuild(Relation heap, Relation index, IndexInfo *indexInfo)
 	/*
 	 * Initialize our state, including the deformed tuple state.
 	 */
-	revmap = brinRevmapInitialize(index, &pagesPerRange);
+	revmap = brinRevmapInitialize(index, &pagesPerRange, NULL);
 	state = initialize_brin_buildstate(index, revmap, pagesPerRange);
 
 	/*
@@ -1045,7 +1047,8 @@ summarize_range(IndexInfo *indexInfo, BrinBuildState *state, Relation heapRel,
 		 * the same.)
 		 */
 		phtup = brinGetTupleForHeapBlock(state->bs_rmAccess, heapBlk, &phbuf,
-										 &offset, &phsz, BUFFER_LOCK_SHARE);
+										 &offset, &phsz, BUFFER_LOCK_SHARE,
+										 NULL);
 		/* the placeholder tuple must exist */
 		if (phtup == NULL)
 			elog(ERROR, "missing placeholder tuple");
@@ -1080,7 +1083,7 @@ brinsummarize(Relation index, Relation heapRel, double *numSummarized,
 	BlockNumber pagesPerRange;
 	Buffer		buf;
 
-	revmap = brinRevmapInitialize(index, &pagesPerRange);
+	revmap = brinRevmapInitialize(index, &pagesPerRange, NULL);
 
 	/*
 	 * Scan the revmap to find unsummarized items.
@@ -1095,7 +1098,7 @@ brinsummarize(Relation index, Relation heapRel, double *numSummarized,
 		CHECK_FOR_INTERRUPTS();
 
 		tup = brinGetTupleForHeapBlock(revmap, heapBlk, &buf, &off, NULL,
-									   BUFFER_LOCK_SHARE);
+									   BUFFER_LOCK_SHARE, NULL);
 		if (tup == NULL)
 		{
 			/* no revmap entry for this heap range. Summarize it. */

src/backend/access/brin/brin_revmap.c

@@ -68,7 +68,8 @@ static void revmap_physical_extend(BrinRevmap *revmap);
  * brinRevmapTerminate when caller is done with it.
  */
 BrinRevmap *
-brinRevmapInitialize(Relation idxrel, BlockNumber *pagesPerRange)
+brinRevmapInitialize(Relation idxrel, BlockNumber *pagesPerRange,
+					 Snapshot snapshot)
 {
 	BrinRevmap *revmap;
 	Buffer		meta;
@@ -77,7 +78,7 @@ brinRevmapInitialize(Relation idxrel, BlockNumber *pagesPerRange)
 
 	meta = ReadBuffer(idxrel, BRIN_METAPAGE_BLKNO);
 	LockBuffer(meta, BUFFER_LOCK_SHARE);
-	page = BufferGetPage(meta, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+	page = BufferGetPage(meta, snapshot, idxrel, BGP_TEST_FOR_OLD_SNAPSHOT);
 	metadata = (BrinMetaPageData *) PageGetContents(page);
 
 	revmap = palloc(sizeof(BrinRevmap));
@@ -187,7 +188,8 @@ brinSetHeapBlockItemptr(Buffer buf, BlockNumber pagesPerRange,
  */
 BrinTuple *
 brinGetTupleForHeapBlock(BrinRevmap *revmap, BlockNumber heapBlk,
-						 Buffer *buf, OffsetNumber *off, Size *size, int mode)
+						 Buffer *buf, OffsetNumber *off, Size *size, int mode,
+						 Snapshot snapshot)
 {
 	Relation	idxRel = revmap->rm_irel;
 	BlockNumber mapBlk;
@@ -264,7 +266,8 @@ brinGetTupleForHeapBlock(BrinRevmap *revmap, BlockNumber heapBlk,
 			*buf = ReadBuffer(idxRel, blk);
 		}
 		LockBuffer(*buf, mode);
-		page = BufferGetPage(*buf, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(*buf, snapshot, idxRel,
+							 BGP_TEST_FOR_OLD_SNAPSHOT);
 
 		/* If we land on a revmap page, start over */
 		if (BRIN_IS_REGULAR_PAGE(page))

src/backend/access/gin/ginbtree.c

@@ -71,7 +71,7 @@ ginTraverseLock(Buffer buffer, bool searchMode)
  * is share-locked, and stack->parent is NULL.
  */
 GinBtreeStack *
-ginFindLeafPage(GinBtree btree, bool searchMode)
+ginFindLeafPage(GinBtree btree, bool searchMode, Snapshot snapshot)
 {
 	GinBtreeStack *stack;
 
@@ -89,7 +89,8 @@ ginFindLeafPage(GinBtree btree, bool searchMode)
 
 		stack->off = InvalidOffsetNumber;
 
-		page = BufferGetPage(stack->buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(stack->buffer, snapshot, btree->index,
+							 BGP_TEST_FOR_OLD_SNAPSHOT);
 
 		access = ginTraverseLock(stack->buffer, searchMode);
 
@@ -115,8 +116,8 @@ ginFindLeafPage(GinBtree btree, bool searchMode)
 
 			stack->buffer = ginStepRight(stack->buffer, btree->index, access);
 			stack->blkno = rightlink;
-			page = BufferGetPage(stack->buffer, NULL, NULL,
-								 BGP_NO_SNAPSHOT_TEST);
+			page = BufferGetPage(stack->buffer, snapshot, btree->index,
+								 BGP_TEST_FOR_OLD_SNAPSHOT);
 
 			if (!searchMode && GinPageIsIncompleteSplit(page))
 				ginFinishSplit(btree, stack, false, NULL);

src/backend/access/gin/gindatapage.c

@@ -1820,7 +1820,7 @@ ginInsertItemPointers(Relation index, BlockNumber rootBlkno,
 	{
 		/* search for the leaf page where the first item should go to */
 		btree.itemptr = insertdata.items[insertdata.curitem];
-		stack = ginFindLeafPage(&btree, false);
+		stack = ginFindLeafPage(&btree, false, NULL);
 
 		ginInsertValue(&btree, stack, &insertdata, buildStats);
 	}
@@ -1830,15 +1830,16 @@ ginInsertItemPointers(Relation index, BlockNumber rootBlkno,
  * Starts a new scan on a posting tree.
  */
 GinBtreeStack *
-ginScanBeginPostingTree(GinBtree btree, Relation index, BlockNumber rootBlkno)
+ginScanBeginPostingTree(GinBtree btree, Relation index, BlockNumber rootBlkno,
+						Snapshot snapshot)
 {
 	GinBtreeStack *stack;
 
 	ginPrepareDataScan(btree, index, rootBlkno);
 
 	btree->fullScan = TRUE;
 
-	stack = ginFindLeafPage(btree, TRUE);
+	stack = ginFindLeafPage(btree, TRUE, snapshot);
 
 	return stack;
 }

src/backend/access/gin/ginget.c

@@ -73,7 +73,7 @@ scanPostingTree(Relation index, GinScanEntry scanEntry,
 	Page		page;
 
 	/* Descend to the leftmost leaf page */
-	stack = ginScanBeginPostingTree(&btree, index, rootPostingTree);
+	stack = ginScanBeginPostingTree(&btree, index, rootPostingTree, snapshot);
 	buffer = stack->buffer;
 	IncrBufferRefCount(buffer); /* prevent unpin in freeGinBtreeStack */
 
@@ -146,7 +146,8 @@ collectMatchBitmap(GinBtreeData *btree, GinBtreeStack *stack,
 		if (moveRightIfItNeeded(btree, stack) == false)
 			return true;
 
-		page = BufferGetPage(stack->buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(stack->buffer, snapshot, btree->index,
+							 BGP_TEST_FOR_OLD_SNAPSHOT);
 		itup = (IndexTuple) PageGetItem(page, PageGetItemId(page, stack->off));
 
 		/*
@@ -320,7 +321,7 @@ startScanEntry(GinState *ginstate, GinScanEntry entry, Snapshot snapshot)
 	ginPrepareEntryScan(&btreeEntry, entry->attnum,
 						entry->queryKey, entry->queryCategory,
 						ginstate);
-	stackEntry = ginFindLeafPage(&btreeEntry, true);
+	stackEntry = ginFindLeafPage(&btreeEntry, true, snapshot);
 	page = BufferGetPage(stackEntry->buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
 	needUnlock = TRUE;
 
@@ -385,7 +386,7 @@ startScanEntry(GinState *ginstate, GinScanEntry entry, Snapshot snapshot)
 			needUnlock = FALSE;
 
 			stack = ginScanBeginPostingTree(&entry->btree, ginstate->index,
-											rootPostingTree);
+											rootPostingTree, snapshot);
 			entry->buffer = stack->buffer;
 
 			/*
@@ -627,7 +628,7 @@ entryLoadMoreItems(GinState *ginstate, GinScanEntry entry,
 			entry->btree.itemptr.ip_posid++;
 		}
 		entry->btree.fullScan = false;
-		stack = ginFindLeafPage(&entry->btree, true);
+		stack = ginFindLeafPage(&entry->btree, true, snapshot);
 
 		/* we don't need the stack, just the buffer. */
 		entry->buffer = stack->buffer;
@@ -1335,8 +1336,8 @@ scanGetCandidate(IndexScanDesc scan, pendingPosition *pos)
 	ItemPointerSetInvalid(&pos->item);
 	for (;;)
 	{
-		page = BufferGetPage(pos->pendingBuffer, NULL,
-							 NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(pos->pendingBuffer, scan->xs_snapshot,
+							 scan->indexRelation, BGP_TEST_FOR_OLD_SNAPSHOT);
 
 		maxoff = PageGetMaxOffsetNumber(page);
 		if (pos->firstOffset > maxoff)
@@ -1516,8 +1517,8 @@ collectMatchesForHeapRow(IndexScanDesc scan, pendingPosition *pos)
 		memset(datumExtracted + pos->firstOffset - 1, 0,
 			   sizeof(bool) * (pos->lastOffset - pos->firstOffset));
 
-		page = BufferGetPage(pos->pendingBuffer, NULL,
-							 NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(pos->pendingBuffer, scan->xs_snapshot,
+							 scan->indexRelation, BGP_TEST_FOR_OLD_SNAPSHOT);
 
 		for (i = 0; i < so->nkeys; i++)
 		{
@@ -1710,7 +1711,8 @@ scanPendingInsert(IndexScanDesc scan, TIDBitmap *tbm, int64 *ntids)
 	*ntids = 0;
 
 	LockBuffer(metabuffer, GIN_SHARE);
-	page = BufferGetPage(metabuffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+	page = BufferGetPage(metabuffer, scan->xs_snapshot, scan->indexRelation,
+						 BGP_TEST_FOR_OLD_SNAPSHOT);
 	blkno = GinPageGetMeta(page)->head;
 
 	/*

src/backend/access/gin/gininsert.c

@@ -192,7 +192,7 @@ ginEntryInsert(GinState *ginstate,
 
 	ginPrepareEntryScan(&btree, attnum, key, category, ginstate);
 
-	stack = ginFindLeafPage(&btree, false);
+	stack = ginFindLeafPage(&btree, false, NULL);
 	page = BufferGetPage(stack->buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
 
 	if (btree.findItem(&btree, stack))

src/backend/access/gist/gistget.c

@@ -336,7 +336,7 @@ gistScanPage(IndexScanDesc scan, GISTSearchItem *pageItem, double *myDistances,
 	buffer = ReadBuffer(scan->indexRelation, pageItem->blkno);
 	LockBuffer(buffer, GIST_SHARE);
 	gistcheckpage(scan->indexRelation, buffer);
-	page = BufferGetPage(buffer, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+	page = BufferGetPage(buffer, scan->xs_snapshot, r, BGP_TEST_FOR_OLD_SNAPSHOT);
 	opaque = GistPageGetOpaque(page);
 
 	/*

src/backend/access/hash/hash.c

@@ -278,7 +278,8 @@ hashgettuple(IndexScanDesc scan, ScanDirection dir)
 
 		buf = so->hashso_curbuf;
 		Assert(BufferIsValid(buf));
-		page = BufferGetPage(buf, NULL, NULL, BGP_NO_SNAPSHOT_TEST);
+		page = BufferGetPage(buf, scan->xs_snapshot, rel,
+							 BGP_TEST_FOR_OLD_SNAPSHOT);
 		maxoffnum = PageGetMaxOffsetNumber(page);
 		for (offnum = ItemPointerGetOffsetNumber(current);
 			 offnum <= maxoffnum;

src/backend/access/hash/hashsearch.c

@@ -188,8 +188,8 @@ _hash_first(IndexScanDesc scan, ScanDirection dir)
 
 	/* Read the metapage */
 	metabuf = _hash_getbuf(rel, HASH_METAPAGE, HASH_READ, LH_META_PAGE);
-	page = BufferGetPage(metabuf, NULL, NULL,
-						 BGP_NO_SNAPSHOT_TEST);
+	page = BufferGetPage(metabuf, scan->xs_snapshot, rel,
+						 BGP_TEST_FOR_OLD_SNAPSHOT);
 	metap = HashPageGetMeta(page);
 
 	/*
@@ -242,8 +242,8 @@ _hash_first(IndexScanDesc scan, ScanDirection dir)
 
 	/* Fetch the primary bucket page for the bucket */
 	buf = _hash_getbuf(rel, blkno, HASH_READ, LH_BUCKET_PAGE);
-	page = BufferGetPage(buf, NULL, NULL,
-						 BGP_NO_SNAPSHOT_TEST);
+	page = BufferGetPage(buf, scan->xs_snapshot, rel,
+						 BGP_TEST_FOR_OLD_SNAPSHOT);
 	opaque = (HashPageOpaque) PageGetSpecialPointer(page);
 	Assert(opaque->hasho_bucket == bucket);
 
@@ -350,6 +350,7 @@ _hash_step(IndexScanDesc scan, Buffer *bufP, ScanDirection dir)
 					_hash_readnext(rel, &buf, &page, &opaque);
 					if (BufferIsValid(buf))
 					{
+						TestForOldSnapshot(scan->xs_snapshot, rel, page);
 						maxoff = PageGetMaxOffsetNumber(page);
 						offnum = _hash_binsearch(page, so->hashso_sk_hash);
 					}
@@ -391,6 +392,7 @@ _hash_step(IndexScanDesc scan, Buffer *bufP, ScanDirection dir)
 					_hash_readprev(rel, &buf, &page, &opaque);
 					if (BufferIsValid(buf))
 					{
+						TestForOldSnapshot(scan->xs_snapshot, rel, page);
 						maxoff = PageGetMaxOffsetNumber(page);
 						offnum = _hash_binsearch_last(page, so->hashso_sk_hash);
 					}