vacuum.c

PostgreSQL 8.1.4的源码 适用于Linux下的开源数据库系统

	 */
	if (use_own_xacts)
	{
		/* matches the StartTransaction in PostgresMain() */
		CommitTransactionCommand();
	}

	/* Turn vacuum cost accounting on or off */
	PG_TRY();
	{
		ListCell   *cur;

		VacuumCostActive = (VacuumCostDelay > 0);
		VacuumCostBalance = 0;

		/*
		 * Loop to process each selected relation.
		 */
		foreach(cur, relations)
		{
			Oid			relid = lfirst_oid(cur);

			if (vacstmt->vacuum)
			{
				if (!vacuum_rel(relid, vacstmt, RELKIND_RELATION))
					all_rels = false;	/* forget about updating dbstats */
			}
			if (vacstmt->analyze)
			{
				MemoryContext old_context = NULL;

				/*
				 * If using separate xacts, start one for analyze. Otherwise,
				 * we can use the outer transaction, but we still need to call
				 * analyze_rel in a memory context that will be cleaned up on
				 * return (else we leak memory while processing multiple
				 * tables).
				 */
				if (use_own_xacts)
				{
					StartTransactionCommand();
					/* functions in indexes may want a snapshot set */
					ActiveSnapshot = CopySnapshot(GetTransactionSnapshot());
				}
				else
					old_context = MemoryContextSwitchTo(anl_context);

				/*
				 * Tell the buffer replacement strategy that vacuum is causing
				 * the IO
				 */
				StrategyHintVacuum(true);

				analyze_rel(relid, vacstmt);

				StrategyHintVacuum(false);

				if (use_own_xacts)
					CommitTransactionCommand();
				else
				{
					MemoryContextSwitchTo(old_context);
					MemoryContextResetAndDeleteChildren(anl_context);
				}
			}
		}
	}
	PG_CATCH();
	{
		/* Make sure cost accounting is turned off after error */
		VacuumCostActive = false;
		PG_RE_THROW();
	}
	PG_END_TRY();

	/* Turn off vacuum cost accounting */
	VacuumCostActive = false;

	/*
	 * Finish up processing.
	 */
	if (use_own_xacts)
	{
		/* here, we are not in a transaction */

		/*
		 * This matches the CommitTransaction waiting for us in
		 * PostgresMain().
		 */
		StartTransactionCommand();
		/*
		 * Re-establish the transaction snapshot.  This is wasted effort
		 * when we are called as a normal utility command, because the
		 * new transaction will be dropped immediately by PostgresMain();
		 * but it's necessary if we are called from autovacuum because
		 * autovacuum might continue on to do an ANALYZE-only call.
		 */
		ActiveSnapshot = CopySnapshot(GetTransactionSnapshot());
	}

	if (vacstmt->vacuum)
	{
		/*
		 * If it was a database-wide VACUUM, print FSM usage statistics (we
		 * don't make you be superuser to see these).
		 */
		if (all_rels)
			PrintFreeSpaceMapStatistics(elevel);

		/*
		 * If we completed a database-wide VACUUM without skipping any
		 * relations, update the database's pg_database row with info about
		 * the transaction IDs used, and try to truncate pg_clog.
		 */
		if (all_rels)
		{
			vac_update_dbstats(MyDatabaseId,
							   initialOldestXmin, initialFreezeLimit);
			vac_truncate_clog(initialOldestXmin, initialFreezeLimit);
		}
	}

	/*
	 * Clean up working storage --- note we must do this after
	 * StartTransactionCommand, else we might be trying to delete the active
	 * context!
	 */
	MemoryContextDelete(vac_context);
	vac_context = NULL;

	if (anl_context)
		MemoryContextDelete(anl_context);
}

/*
 * Build a list of Oids for each relation to be processed
 *
 * The list is built in vac_context so that it will survive across our
 * per-relation transactions.
 */
static List *
get_rel_oids(List *relids, const RangeVar *vacrel, const char *stmttype)
{
	List	   *oid_list = NIL;
	MemoryContext oldcontext;

	/* List supplied by VACUUM's caller? */
	if (relids)
		return relids;

	if (vacrel)
	{
		/* Process a specific relation */
		Oid			relid;

		relid = RangeVarGetRelid(vacrel, false);

		/* Make a relation list entry for this guy */
		oldcontext = MemoryContextSwitchTo(vac_context);
		oid_list = lappend_oid(oid_list, relid);
		MemoryContextSwitchTo(oldcontext);
	}
	else
	{
		/* Process all plain relations listed in pg_class */
		Relation	pgclass;
		HeapScanDesc scan;
		HeapTuple	tuple;
		ScanKeyData key;

		ScanKeyInit(&key,
					Anum_pg_class_relkind,
					BTEqualStrategyNumber, F_CHAREQ,
					CharGetDatum(RELKIND_RELATION));

		pgclass = heap_open(RelationRelationId, AccessShareLock);

		scan = heap_beginscan(pgclass, SnapshotNow, 1, &key);

		while ((tuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
		{
			/* Make a relation list entry for this guy */
			oldcontext = MemoryContextSwitchTo(vac_context);
			oid_list = lappend_oid(oid_list, HeapTupleGetOid(tuple));
			MemoryContextSwitchTo(oldcontext);
		}

		heap_endscan(scan);
		heap_close(pgclass, AccessShareLock);
	}

	return oid_list;
}

/*
 * vacuum_set_xid_limits() -- compute oldest-Xmin and freeze cutoff points
 */
void
vacuum_set_xid_limits(VacuumStmt *vacstmt, bool sharedRel,
					  TransactionId *oldestXmin,
					  TransactionId *freezeLimit)
{
	TransactionId limit;

	*oldestXmin = GetOldestXmin(sharedRel);

	Assert(TransactionIdIsNormal(*oldestXmin));

	if (vacstmt->freeze)
	{
		/* FREEZE option: use oldest Xmin as freeze cutoff too */
		limit = *oldestXmin;
	}
	else
	{
		/*
		 * Normal case: freeze cutoff is well in the past, to wit, about
		 * halfway to the wrap horizon
		 */
		limit = GetCurrentTransactionId() - (MaxTransactionId >> 2);
	}

	/*
	 * Be careful not to generate a "permanent" XID
	 */
	if (!TransactionIdIsNormal(limit))
		limit = FirstNormalTransactionId;

	/*
	 * Ensure sane relationship of limits
	 */
	if (TransactionIdFollows(limit, *oldestXmin))
	{
		ereport(WARNING,
				(errmsg("oldest xmin is far in the past"),
				 errhint("Close open transactions soon to avoid wraparound problems.")));
		limit = *oldestXmin;
	}

	*freezeLimit = limit;
}


/*
 *	vac_update_relstats() -- update statistics for one relation
 *
 *		Update the whole-relation statistics that are kept in its pg_class
 *		row.  There are additional stats that will be updated if we are
 *		doing ANALYZE, but we always update these stats.  This routine works
 *		for both index and heap relation entries in pg_class.
 *
 *		We violate no-overwrite semantics here by storing new values for the
 *		statistics columns directly into the pg_class tuple that's already on
 *		the page.  The reason for this is that if we updated these tuples in
 *		the usual way, vacuuming pg_class itself wouldn't work very well ---
 *		by the time we got done with a vacuum cycle, most of the tuples in
 *		pg_class would've been obsoleted.  Of course, this only works for
 *		fixed-size never-null columns, but these are.
 *
 *		This routine is shared by full VACUUM, lazy VACUUM, and stand-alone
 *		ANALYZE.
 */
void
vac_update_relstats(Oid relid, BlockNumber num_pages, double num_tuples,
					bool hasindex)
{
	Relation	rd;
	HeapTupleData rtup;
	HeapTuple	ctup;
	Form_pg_class pgcform;
	Buffer		buffer;

	/*
	 * update number of tuples and number of pages in pg_class
	 */
	rd = heap_open(RelationRelationId, RowExclusiveLock);

	ctup = SearchSysCache(RELOID,
						  ObjectIdGetDatum(relid),
						  0, 0, 0);
	if (!HeapTupleIsValid(ctup))
		elog(ERROR, "pg_class entry for relid %u vanished during vacuuming",
			 relid);

	/* get the buffer cache tuple */
	rtup.t_self = ctup->t_self;
	ReleaseSysCache(ctup);
	if (!heap_fetch(rd, SnapshotNow, &rtup, &buffer, false, NULL))
		elog(ERROR, "pg_class entry for relid %u vanished during vacuuming",
			 relid);

	/* ensure no one else does this at the same time */
	LockBuffer(buffer, BUFFER_LOCK_EXCLUSIVE);

	/* overwrite the existing statistics in the tuple */
	pgcform = (Form_pg_class) GETSTRUCT(&rtup);
	pgcform->relpages = (int32) num_pages;
	pgcform->reltuples = (float4) num_tuples;
	pgcform->relhasindex = hasindex;

	/*
	 * If we have discovered that there are no indexes, then there's no
	 * primary key either.	This could be done more thoroughly...
	 */
	if (!hasindex)
		pgcform->relhaspkey = false;

	LockBuffer(buffer, BUFFER_LOCK_UNLOCK);

	/*
	 * Invalidate the tuple in the catcaches; this also arranges to flush the
	 * relation's relcache entry.  (If we fail to commit for some reason, no
	 * flush will occur, but no great harm is done since there are no
	 * noncritical state updates here.)
	 */
	CacheInvalidateHeapTuple(rd, &rtup);

	/* Write the buffer */
	WriteBuffer(buffer);

	heap_close(rd, RowExclusiveLock);
}


/*
 *	vac_update_dbstats() -- update statistics for one database
 *
 *		Update the whole-database statistics that are kept in its pg_database
 *		row, and the flat-file copy of pg_database.
 *
 *		We violate no-overwrite semantics here by storing new values for the
 *		statistics columns directly into the tuple that's already on the page.
 *		As with vac_update_relstats, this avoids leaving dead tuples behind
 *		after a VACUUM.
 *
 *		This routine is shared by full and lazy VACUUM.  Note that it is only
 *		applied after a database-wide VACUUM operation.
 */
static void
vac_update_dbstats(Oid dbid,
				   TransactionId vacuumXID,
				   TransactionId frozenXID)
{
	Relation	relation;
	ScanKeyData entry[1];
	HeapScanDesc scan;
	HeapTuple	tuple;
	Form_pg_database dbform;

	relation = heap_open(DatabaseRelationId, RowExclusiveLock);

	/* Must use a heap scan, since there's no syscache for pg_database */
	ScanKeyInit(&entry[0],
				ObjectIdAttributeNumber,
				BTEqualStrategyNumber, F_OIDEQ,
				ObjectIdGetDatum(dbid));

	scan = heap_beginscan(relation, SnapshotNow, 1, entry);

	tuple = heap_getnext(scan, ForwardScanDirection);

	if (!HeapTupleIsValid(tuple))
		elog(ERROR, "could not find tuple for database %u", dbid);

	/* ensure no one else does this at the same time */
	LockBuffer(scan->rs_cbuf, BUFFER_LOCK_EXCLUSIVE);

	dbform = (Form_pg_database) GETSTRUCT(tuple);

	/* overwrite the existing statistics in the tuple */
	dbform->datvacuumxid = vacuumXID;
	dbform->datfrozenxid = frozenXID;

	LockBuffer(scan->rs_cbuf, BUFFER_LOCK_UNLOCK);

	/* invalidate the tuple in the cache and write the buffer */
	CacheInvalidateHeapTuple(relation, tuple);
	WriteNoReleaseBuffer(scan->rs_cbuf);

	heap_endscan(scan);

	heap_close(relation, RowExclusiveLock);

	/* Mark the flat-file copy of pg_database for update at commit */
	database_file_update_needed();
}


/*
 *	vac_truncate_clog() -- attempt to truncate the commit log
 *
 *		Scan pg_database to determine the system-wide oldest datvacuumxid,
 *		and use it to truncate the transaction commit log (pg_clog).
 *		Also update the XID wrap limit point maintained by varsup.c.
 *
 *		We also generate a warning if the system-wide oldest datfrozenxid
 *		seems to be in danger of wrapping around.  This is a long-in-advance
 *		warning; if we start getting uncomfortably close, GetNewTransactionId
 *		will generate more-annoying warnings, and ultimately refuse to issue
 *		any more new XIDs.
 *
 *		The passed XIDs are simply the ones I just wrote into my pg_database
 *		entry.	They're used to initialize the "min" calculations.
 *
 *		This routine is shared by full and lazy VACUUM.  Note that it is only
 *		applied after a database-wide VACUUM operation.
 */
static void
vac_truncate_clog(TransactionId vacuumXID, TransactionId frozenXID)
{
	TransactionId myXID = GetCurrentTransactionId();
	Relation	relation;
	HeapScanDesc scan;
	HeapTuple	tuple;
	int32		age;
	NameData	oldest_datname;
	bool		vacuumAlreadyWrapped = false;
	bool		frozenAlreadyWrapped = false;

	/* init oldest_datname to sync with my frozenXID */
	namestrcpy(&oldest_datname, get_database_name(MyDatabaseId));

	/*
	 * Note: the "already wrapped" cases should now be impossible due to the