提交 aa21da20 编写于 作者: T Tom Lane

Put back BufferSync() for WAL after database creation.

Also, add some comments about what it's for...
上级 52aa7203
......@@ -8,7 +8,7 @@
*
*
* IDENTIFICATION
* $Header: /cvsroot/pgsql/src/backend/commands/dbcommands.c,v 1.68 2000/11/16 22:30:18 tgl Exp $
* $Header: /cvsroot/pgsql/src/backend/commands/dbcommands.c,v 1.69 2000/11/18 03:36:48 tgl Exp $
*
*-------------------------------------------------------------------------
*/
......@@ -154,7 +154,10 @@ createdb(const char *dbname, const char *dbpath,
/* ... otherwise we'd be open to shell exploits below */
#ifdef XLOG
/* Try to force any dirty buffers out to disk */
/* Force dirty buffers out to disk, to ensure source database is
* up-to-date for the copy. (We really only need to flush buffers
* for the source database...)
*/
BufferSync();
#endif
......@@ -251,6 +254,14 @@ createdb(const char *dbname, const char *dbpath,
/* Close pg_database, but keep lock till commit */
heap_close(pg_database_rel, NoLock);
#ifdef XLOG
/* Force dirty buffers out to disk, so that newly-connecting backends
* will see the new database in pg_database right away. (They'll see
* an uncommitted tuple, but they don't care; see GetRawDatabaseInfo.)
*/
BufferSync();
#endif
}
......
Markdown is supported
0% .
You are about to add 0 people to the discussion. Proceed with caution.
先完成此消息的编辑!
想要评论请 注册