Add new block-by-block strategy for CREATE DATABASE.

Because this strategy logs changes on a block-by-block basis, it
avoids the need to checkpoint before and after the operation.
However, because it logs each changed block individually, it might
generate a lot of extra write-ahead logging if the template database
is large. Therefore, the older strategy remains available via a new
STRATEGY parameter to CREATE DATABASE, and a corresponding --strategy
option to createdb.

Somewhat controversially, this patch assembles the list of relations
to be copied to the new database by reading the pg_class relation of
the template database. Cross-database access like this isn't normally
possible, but it can be made to work here because there can't be any
connections to the database being copied, nor can it contain any
in-doubt transactions. Even so, we have to use lower-level interfaces
than normal, since the table scan and relcache interfaces will not
work for a database to which we're not connected. The advantage of
this approach is that we do not need to rely on the filesystem to
determine what ought to be copied, but instead on PostgreSQL's own
knowledge of the database structure. This avoids, for example,
copying stray files that happen to be located in the source database
directory.

Dilip Kumar, with a fairly large number of cosmetic changes by me.
Reviewed and tested by Ashutosh Sharma, Andres Freund, John Naylor,
Greg Nancarrow, Neha Sharma. Additional feedback from Bruce Momjian,
Heikki Linnakangas, Julien Rouhaud, Adam Brusselback, Kyotaro
Horiguchi, Tomas Vondra, Andrew Dunstan, Álvaro Herrera, and others.

Discussion: http://postgr.es/m/CA+TgmoYtcdxBjLh31DLxUXHxFVMPGzrU5_T=CYCvRyFHywSBUQ@mail.gmail.com

Author: robertmhaas

contrib/bloom/blinsert.c

@@ -173,7 +173,7 @@ blbuildempty(Relation index)
 	 * Write the page and log it.  It might seem that an immediate sync would
 	 * be sufficient to guarantee that the file exists on disk, but recovery
 	 * itself might remove it while replaying, for example, an
-	 * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record.  Therefore, we need
+	 * XLOG_DBASE_CREATE* or XLOG_TBLSPC_CREATE record.  Therefore, we need
 	 * this even when wal_level=minimal.
 	 */
 	PageSetChecksumInplace(metapage, BLOOM_METAPAGE_BLKNO);

doc/src/sgml/monitoring.sgml

@@ -1502,6 +1502,10 @@ postgres   27093  0.0  0.0  30096  2752 ?        Ss   11:34   0:00 postgres: ser
       <entry><literal>TwophaseFileWrite</literal></entry>
       <entry>Waiting for a write of a two phase state file.</entry>
      </row>
+     <row>
+      <entry><literal>VersionFileWrite</literal></entry>
+      <entry>Waiting for the version file to be written while creating a database.</entry>
+     </row>
      <row>
       <entry><literal>WALBootstrapSync</literal></entry>
       <entry>Waiting for WAL to reach durable storage during

doc/src/sgml/ref/create_database.sgml

@@ -25,6 +25,7 @@ CREATE DATABASE <replaceable class="parameter">name</replaceable>
     [ [ WITH ] [ OWNER [=] <replaceable class="parameter">user_name</replaceable> ]
            [ TEMPLATE [=] <replaceable class="parameter">template</replaceable> ]
            [ ENCODING [=] <replaceable class="parameter">encoding</replaceable> ]
+           [ STRATEGY [=] <replaceable class="parameter">strategy</replaceable> ] ]
            [ LOCALE [=] <replaceable class="parameter">locale</replaceable> ]
            [ LC_COLLATE [=] <replaceable class="parameter">lc_collate</replaceable> ]
            [ LC_CTYPE [=] <replaceable class="parameter">lc_ctype</replaceable> ]
@@ -118,6 +119,27 @@ CREATE DATABASE <replaceable class="parameter">name</replaceable>
        </para>
       </listitem>
      </varlistentry>
+     <varlistentry id="create-database-strategy" xreflabel="CREATE DATABASE STRATEGY">
+      <term><replaceable class="parameter">strategy</replaceable></term>
+      <listitem>
+       <para>
+        Strategy to be used in creating the new database.  If
+        the <literal>WAL_LOG</literal> strategy is used, the database will be
+        copied block by block and each block will be separately written
+        to the write-ahead log. This is the most efficient strategy in
+        cases where the template database is small, and therefore it is the
+        default. The older <literal>FILE_COPY</literal> strategy is also
+        available. This strategy writes a small record to the write-ahead log
+        for each tablespace used by the target database. Each such record
+        represents copying an entire directory to a new location at the
+        filesystem level. While this does reduce the write-ahed
+        log volume substantially, especially if the template database is large,
+        it also forces the system to perform a checkpoint both before and
+        after the creation of the new database. In some situations, this may
+        have a noticeable negative impact on overall system performance.
+       </para>
+      </listitem>
+     </varlistentry>
      <varlistentry>
       <term><replaceable class="parameter">locale</replaceable></term>
       <listitem>

doc/src/sgml/ref/createdb.sgml

@@ -177,6 +177,17 @@ PostgreSQL documentation
       </listitem>
      </varlistentry>
 
+     <varlistentry>
+      <term><option>-S <replaceable class="parameter">template</replaceable></option></term>
+      <term><option>--strategy=<replaceable class="parameter">strategy</replaceable></option></term>
+      <listitem>
+       <para>
+        Specifies the database creation strategy.  See
+        <xref linkend="create-database-strategy" /> for more details.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term><option>-T <replaceable class="parameter">template</replaceable></option></term>
       <term><option>--template=<replaceable class="parameter">template</replaceable></option></term>

src/backend/access/heap/heapam_handler.c

@@ -593,15 +593,15 @@ heapam_relation_set_new_filenode(Relation rel,
 	 */
 	*minmulti = GetOldestMultiXactId();
 
-	srel = RelationCreateStorage(*newrnode, persistence);
+	srel = RelationCreateStorage(*newrnode, persistence, true);
 
 	/*
 	 * If required, set up an init fork for an unlogged table so that it can
 	 * be correctly reinitialized on restart.  An immediate sync is required
 	 * even if the page has been logged, because the write did not go through
 	 * shared_buffers and therefore a concurrent checkpoint may have moved the
 	 * redo pointer past our xlog record.  Recovery may as well remove it
-	 * while replaying, for example, XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE
+	 * while replaying, for example, XLOG_DBASE_CREATE* or XLOG_TBLSPC_CREATE
 	 * record. Therefore, logging is necessary even if wal_level=minimal.
 	 */
 	if (persistence == RELPERSISTENCE_UNLOGGED)
@@ -645,7 +645,7 @@ heapam_relation_copy_data(Relation rel, const RelFileNode *newrnode)
 	 * NOTE: any conflict in relfilenode value will be caught in
 	 * RelationCreateStorage().
 	 */
-	RelationCreateStorage(*newrnode, rel->rd_rel->relpersistence);
+	RelationCreateStorage(*newrnode, rel->rd_rel->relpersistence, true);
 
 	/* copy main fork */
 	RelationCopyStorage(RelationGetSmgr(rel), dstrel, MAIN_FORKNUM,

src/backend/access/nbtree/nbtree.c

@@ -161,7 +161,7 @@ btbuildempty(Relation index)
 	 * Write the page and log it.  It might seem that an immediate sync would
 	 * be sufficient to guarantee that the file exists on disk, but recovery
 	 * itself might remove it while replaying, for example, an
-	 * XLOG_DBASE_CREATE or XLOG_TBLSPC_CREATE record.  Therefore, we need
+	 * XLOG_DBASE_CREATE* or XLOG_TBLSPC_CREATE record.  Therefore, we need
 	 * this even when wal_level=minimal.
 	 */
 	PageSetChecksumInplace(metapage, BTREE_METAPAGE);

src/backend/access/rmgrdesc/dbasedesc.c

@@ -24,14 +24,23 @@ dbase_desc(StringInfo buf, XLogReaderState *record)
 	char	   *rec = XLogRecGetData(record);
 	uint8		info = XLogRecGetInfo(record) & ~XLR_INFO_MASK;
 
-	if (info == XLOG_DBASE_CREATE)
+	if (info == XLOG_DBASE_CREATE_FILE_COPY)
 	{
-		xl_dbase_create_rec *xlrec = (xl_dbase_create_rec *) rec;
+		xl_dbase_create_file_copy_rec *xlrec =
+		(xl_dbase_create_file_copy_rec *) rec;
 
 		appendStringInfo(buf, "copy dir %u/%u to %u/%u",
 						 xlrec->src_tablespace_id, xlrec->src_db_id,
 						 xlrec->tablespace_id, xlrec->db_id);
 	}
+	else if (info == XLOG_DBASE_CREATE_WAL_LOG)
+	{
+		xl_dbase_create_wal_log_rec *xlrec =
+		(xl_dbase_create_wal_log_rec *) rec;
+
+		appendStringInfo(buf, "create dir %u/%u",
+						 xlrec->tablespace_id, xlrec->db_id);
+	}
 	else if (info == XLOG_DBASE_DROP)
 	{
 		xl_dbase_drop_rec *xlrec = (xl_dbase_drop_rec *) rec;
@@ -51,8 +60,11 @@ dbase_identify(uint8 info)
 
 	switch (info & ~XLR_INFO_MASK)
 	{
-		case XLOG_DBASE_CREATE:
-			id = "CREATE";
+		case XLOG_DBASE_CREATE_FILE_COPY:
+			id = "CREATE_FILE_COPY";
+			break;
+		case XLOG_DBASE_CREATE_WAL_LOG:
+			id = "CREATE_WAL_LOG";
 			break;
 		case XLOG_DBASE_DROP:
 			id = "DROP";

src/backend/access/transam/xlogutils.c

@@ -484,7 +484,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
 	{
 		/* page exists in file */
 		buffer = ReadBufferWithoutRelcache(rnode, forknum, blkno,
-										   mode, NULL);
+										   mode, NULL, true);
 	}
 	else
 	{
@@ -509,7 +509,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
 				ReleaseBuffer(buffer);
 			}
 			buffer = ReadBufferWithoutRelcache(rnode, forknum,
-											   P_NEW, mode, NULL);
+											   P_NEW, mode, NULL, true);
 		}
 		while (BufferGetBlockNumber(buffer) < blkno);
 		/* Handle the corner case that P_NEW returns non-consecutive pages */
@@ -519,7 +519,7 @@ XLogReadBufferExtended(RelFileNode rnode, ForkNumber forknum,
 				LockBuffer(buffer, BUFFER_LOCK_UNLOCK);
 			ReleaseBuffer(buffer);
 			buffer = ReadBufferWithoutRelcache(rnode, forknum, blkno,
-											   mode, NULL);
+											   mode, NULL, true);
 		}
 	}
 

src/backend/catalog/heap.c

@@ -387,7 +387,7 @@ heap_create(const char *relname,
 											relpersistence,
 											relfrozenxid, relminmxid);
 		else if (RELKIND_HAS_STORAGE(rel->rd_rel->relkind))
-			RelationCreateStorage(rel->rd_node, relpersistence);
+			RelationCreateStorage(rel->rd_node, relpersistence, true);
 		else
 			Assert(false);
 	}

src/backend/catalog/storage.c

@@ -112,12 +112,14 @@ AddPendingSync(const RelFileNode *rnode)
  * modules that need them.
  *
  * This function is transactional. The creation is WAL-logged, and if the
- * transaction aborts later on, the storage will be destroyed.
+ * transaction aborts later on, the storage will be destroyed.  A caller
+ * that does not want the storage to be destroyed in case of an abort may
+ * pass register_delete = false.
  */
 SMgrRelation
-RelationCreateStorage(RelFileNode rnode, char relpersistence)
+RelationCreateStorage(RelFileNode rnode, char relpersistence,
+					  bool register_delete)
 {
-	PendingRelDelete *pending;
 	SMgrRelation srel;
 	BackendId	backend;
 	bool		needs_wal;
@@ -149,15 +151,23 @@ RelationCreateStorage(RelFileNode rnode, char relpersistence)
 	if (needs_wal)
 		log_smgrcreate(&srel->smgr_rnode.node, MAIN_FORKNUM);
 
-	/* Add the relation to the list of stuff to delete at abort */
-	pending = (PendingRelDelete *)
-		MemoryContextAlloc(TopMemoryContext, sizeof(PendingRelDelete));
-	pending->relnode = rnode;
-	pending->backend = backend;
-	pending->atCommit = false;	/* delete if abort */
-	pending->nestLevel = GetCurrentTransactionNestLevel();
-	pending->next = pendingDeletes;
-	pendingDeletes = pending;
+	/*
+	 * Add the relation to the list of stuff to delete at abort, if we are
+	 * asked to do so.
+	 */
+	if (register_delete)
+	{
+		PendingRelDelete *pending;
+
+		pending = (PendingRelDelete *)
+			MemoryContextAlloc(TopMemoryContext, sizeof(PendingRelDelete));
+		pending->relnode = rnode;
+		pending->backend = backend;
+		pending->atCommit = false;	/* delete if abort */
+		pending->nestLevel = GetCurrentTransactionNestLevel();
+		pending->next = pendingDeletes;
+		pendingDeletes = pending;
+	}
 
 	if (relpersistence == RELPERSISTENCE_PERMANENT && !XLogIsNeeded())
 	{