Force default wal_sync_method to be fdatasync on Linux.

Recent versions of the Linux system header files cause xlogdefs.h to
believe that open_datasync should be the default sync method, whereas
formerly fdatasync was the default on Linux.  open_datasync is a bad
choice, first because it doesn't actually outperform fdatasync (in fact
the reverse), and second because we try to use O_DIRECT with it, causing
failures on certain filesystems (e.g., ext4 with data=journal option).
This part of the patch is largely per a proposal from Marti Raudsepp.
More extensive changes are likely to follow in HEAD, but this is as much
change as we want to back-patch.

Also clean up confusing code and incorrect documentation surrounding the
fsync_writethrough option.  Those changes shouldn't result in any actual
behavioral change, but I chose to back-patch them anyway to keep the
branches looking similar in this area.

In 9.0 and HEAD, also do some copy-editing on the WAL Reliability
documentation section.

Back-patch to all supported branches, since any of them might get used
on modern Linux versions.

doc/src/sgml/config.sgml

@@ -1460,18 +1460,19 @@ SET ENABLE_SEQSCAN TO OFF;
        <para>
         While turning off <varname>fsync</varname> is often a performance
         benefit, this can result in unrecoverable data corruption in
-        the event of an unexpected system shutdown or crash.  Thus it
-        is only advisable to turn off  <varname>fsync</varname> if
+        the event of a power failure or system crash.  Thus it
+        is only advisable to turn off <varname>fsync</varname> if
         you can easily recreate your entire database from external
         data.
        </para>
 
        <para>
         Examples of safe circumstances for turning off
-        <varname>fsync</varname> include the initial loading a new
+        <varname>fsync</varname> include the initial loading of a new
         database cluster from a backup file, using a database cluster
-        for processing statistics on an hourly basis which is then
-        recreated, or for a reporting read-only database clone which
+        for processing a batch of data after which the database
+        will be thrown away and recreated,
+        or for a read-only database clone which
         gets recreated frequently and is not used for failover.  High
         quality hardware alone is not a sufficient justification for
         turning off <varname>fsync</varname>.
@@ -1554,12 +1555,12 @@ SET ENABLE_SEQSCAN TO OFF;
         </listitem>
         <listitem>
         <para>
-         <literal>fsync_writethrough</> (call <function>fsync()</> at each commit, forcing write-through of any disk write cache)
+         <literal>fsync</> (call <function>fsync()</> at each commit)
         </para>
         </listitem>
         <listitem>
         <para>
-         <literal>fsync</> (call <function>fsync()</> at each commit)
+         <literal>fsync_writethrough</> (call <function>fsync()</> at each commit, forcing write-through of any disk write cache)
         </para>
         </listitem>
         <listitem>
@@ -1569,16 +1570,15 @@ SET ENABLE_SEQSCAN TO OFF;
         </listitem>
        </itemizedlist>
        <para>
-        Not all of these choices are available on all platforms.
         The <literal>open_</>* options also use <literal>O_DIRECT</> if available.
+        Not all of these choices are available on all platforms.
         The default is the first method in the above list that is supported
-        by the platform.  The default is not necessarily ideal; it might be
+        by the platform, except that <literal>fdatasync</> is the default on
+        Linux.  The default is not necessarily ideal; it might be
         necessary to change this setting or other aspects of your system
         configuration in order to create a crash-safe configuration or
         achieve optimal performance.
         These aspects are discussed in <xref linkend="wal-reliability">.
-        The utility <filename>src/tools/fsync</> in the PostgreSQL source tree
-        can do performance testing of various fsync methods.
         This parameter can only be set in the <filename>postgresql.conf</>
         file or on the server command line.
        </para>
@@ -1686,21 +1686,20 @@ SET ENABLE_SEQSCAN TO OFF;
         When the commit data for a transaction is flushed to disk, any
         additional commits ready at that time are also flushed out.
         <varname>commit_delay</varname> adds a time delay, set in
-        microseconds, before writing some commit records to the WAL
-        buffer and flushing the buffer out to disks. A nonzero delay
-        can allow more transactions to be committed with only one call
-        to the active <varname>wal_sync_method</varname>, if
+        microseconds, before a transaction attempts to
+        flush the WAL buffer out to disk.  A nonzero delay can allow more
+        transactions to be committed with only one flush operation, if
         system load is high enough that additional transactions become
         ready to commit within the given interval. But the delay is
         just wasted if no other transactions become ready to
         commit. Therefore, the delay is only performed if at least
         <varname>commit_siblings</varname> other transactions are
         active at the instant that a server process has written its
-        commit record. The default is zero (no delay).  Since
-        all pending commit data flushes are written at every flush
-        regardless of this setting, it is rare that adding delay to
-        that by increasing this parameter will actually improve commit
-        performance.
+        commit record.
+        The default <varname>commit_delay</> is zero (no delay).
+        Since all pending commit data will be written at every flush
+        regardless of this setting, it is rare that adding delay
+        by increasing this parameter will actually improve performance.
        </para>
       </listitem>
      </varlistentry>

doc/src/sgml/wal.sgml

@@ -27,7 +27,7 @@
   </para>
 
   <para>
-   While forcing data periodically to the disk platters might seem like
+   While forcing data to the disk platters periodically might seem like
    a simple operation, it is not. Because disk drives are dramatically
    slower than main memory and CPUs, several layers of caching exist
    between the computer's main memory and the disk platters.
@@ -48,7 +48,7 @@
    some later time. Such caches can be a reliability hazard because the
    memory in the disk controller cache is volatile, and will lose its
    contents in a power failure.  Better controller cards have
-   <firstterm>battery-backed unit</> (<acronym>BBU</>) caches, meaning
+   <firstterm>battery-backup units</> (<acronym>BBU</>s), meaning
    the card has a battery that
    maintains power to the cache in case of system power loss.  After power
    is restored the data will be written to the disk drives.
@@ -57,15 +57,10 @@
   <para>
    And finally, most disk drives have caches. Some are write-through
    while some are write-back, and the same concerns about data loss
-   exist for write-back drive caches as exist for disk controller
+   exist for write-back drive caches as for disk controller
    caches.  Consumer-grade IDE and SATA drives are particularly likely
-   to have write-back caches that will not survive a power failure,
-   though <acronym>ATAPI-6</> introduced a drive cache flush command
-   (<command>FLUSH CACHE EXT</>) that some file systems use, e.g.
-   <acronym>ZFS</>, <acronym>ext4</>.  (The SCSI command
-   <command>SYNCHRONIZE CACHE</> has long been available.) Many
-   solid-state drives (SSD) also have volatile write-back caches, and
-   many do not honor cache flush commands by default.
+   to have write-back caches that will not survive a power failure.  Many
+   solid-state drives (SSD) also have volatile write-back caches.
   </para>
 
   <para>
@@ -81,7 +76,7 @@
         a <literal>*</> next to <literal>Write cache</>.  <command>hdparm -W</>
         can be used to turn off write caching.  SCSI drives can be queried
         using <ulink url="http://sg.danny.cz/sg/sdparm.html"><application>sdparm</></ulink>.
-        for SCSI drives.  Use <command>sdparm --get=WCE</command> to check
+        Use <command>sdparm --get=WCE</command> to check
         whether the write cache is enabled and <command>sdparm --clear=WCE</>
         to disable it.
       </para>
@@ -107,35 +102,40 @@
     <listitem>
       <para>
         On <productname>Windows</>, if <varname>wal_sync_method</> is
-        <literal>open_datasync</> (the default), write caching is disabled
-        by unchecking <literal>My Computer\Open\{select disk drive}\Properties\Hardware\Properties\Policies\Enable write caching on the disk</>.
-        Alternatively, set <varname>wal_sync_method</varname> to <literal>fsync</> or <literal>fsync_writethrough</>, which never do write caching.
+        <literal>open_datasync</> (the default), write caching can be disabled
+        by unchecking <literal>My Computer\Open\<replaceable>disk drive</>\Properties\Hardware\Properties\Policies\Enable write caching on the disk</>.
+        Alternatively, set <varname>wal_sync_method</varname> to
+        <literal>fsync</> or <literal>fsync_writethrough</>, which prevent
+        write caching.
       </para>
     </listitem>
 
     <listitem>
       <para>
-        On <productname>MacOS X</productname>, write caching can be disabled by
+        On <productname>Mac OS X</productname>, write caching can be prevented by
         setting <varname>wal_sync_method</> to <literal>fsync_writethrough</>.
       </para>
     </listitem>
   </itemizedlist>
 
   <para>
-   Many file systems that use write barriers (e.g.  <acronym>ZFS</>,
-   <acronym>ext4</>) internally use <command>FLUSH CACHE EXT</> or
-   <command>SYNCHRONIZE CACHE</> commands to flush data to the platters on
-   write-back-enabled drives.  Unfortunately, such write barrier file
-   systems behave suboptimally when combined with battery-backed unit
+   Recent SATA drives (those following <acronym>ATAPI-6</> or later)
+   offer a drive cache flush command (<command>FLUSH CACHE EXT</>),
+   while SCSI drives have long supported a similar command
+   <command>SYNCHRONIZE CACHE</>.  These commands are not directly
+   accessible to <productname>PostgreSQL</>, but some file systems
+   (e.g., <acronym>ZFS</>, <acronym>ext4</>) can use them to flush
+   data to the platters on write-back-enabled drives.  Unfortunately, such
+   file systems behave suboptimally when combined with battery-backup unit
    (<acronym>BBU</>) disk controllers.  In such setups, the synchronize
-   command forces all data from the BBU to the disks, eliminating much
-   of the benefit of the BBU.  You can run the utility
+   command forces all data from the controller cache to the disks,
+   eliminating much of the benefit of the BBU.  You can run the utility
    <filename>src/tools/fsync</> in the PostgreSQL source tree to see
    if you are affected.  If you are affected, the performance benefits
-   of the BBU cache can be regained by turning off write barriers in
+   of the BBU can be regained by turning off write barriers in
    the file system or reconfiguring the disk controller, if that is
    an option.  If write barriers are turned off, make sure the battery
-   remains active; a faulty battery can potentially lead to data loss.
+   remains functional; a faulty battery can potentially lead to data loss.
    Hopefully file system and disk controller designers will eventually
    address this suboptimal behavior.
   </para>
@@ -148,6 +148,8 @@
    ensure data integrity.  Avoid disk controllers that have non-battery-backed
    write caches.  At the drive level, disable write-back caching if the
    drive cannot guarantee the data will be written before shutdown.
+   If you use SSDs, be aware that many of these do not honor cache flush
+   commands by default.
    You can test for reliable I/O subsystem behavior using <ulink
    url="http://brad.livejournal.com/2116715.html"><filename>diskchecker.pl</filename></ulink>.
   </para>
@@ -157,16 +159,17 @@
    operations themselves. Disk platters are divided into sectors,
    commonly 512 bytes each.  Every physical read or write operation
    processes a whole sector.
-   When a write request arrives at the drive, it might be for 512 bytes,
-   1024 bytes, or 8192 bytes, and the process of writing could fail due
+   When a write request arrives at the drive, it might be for some multiple
+   of 512 bytes (<productname>PostgreSQL</> typically writes 8192 bytes, or
+   16 sectors, at a time), and the process of writing could fail due
    to power loss at any time, meaning some of the 512-byte sectors were
-   written, and others were not.  To guard against such failures,
+   written while others were not.  To guard against such failures,
    <productname>PostgreSQL</> periodically writes full page images to
    permanent WAL storage <emphasis>before</> modifying the actual page on
    disk. By doing this, during crash recovery <productname>PostgreSQL</> can
-   restore partially-written pages.  If you have a battery-backed disk
+   restore partially-written pages from WAL.  If you have a battery-backed disk
    controller or file-system software that prevents partial page writes
-   (e.g., ZFS),  you can turn off this page imaging by turning off the
+   (e.g., ZFS), you can safely turn off this page imaging by turning off the
    <xref linkend="guc-full-page-writes"> parameter.
   </para>
  </sect1>

src/backend/storage/file/fd.c

@@ -260,12 +260,13 @@ static bool looks_like_temp_rel_name(const char *name);
 int
 pg_fsync(int fd)
 {
-#ifndef HAVE_FSYNC_WRITETHROUGH_ONLY
-	if (sync_method != SYNC_METHOD_FSYNC_WRITETHROUGH)
-		return pg_fsync_no_writethrough(fd);
+	/* #if is to skip the sync_method test if there's no need for it */
+#if defined(HAVE_FSYNC_WRITETHROUGH) && !defined(FSYNC_WRITETHROUGH_IS_FSYNC)
+	if (sync_method == SYNC_METHOD_FSYNC_WRITETHROUGH)
+		return pg_fsync_writethrough(fd);
 	else
 #endif
-		return pg_fsync_writethrough(fd);
+		return pg_fsync_no_writethrough(fd);
 }
 
 

src/backend/utils/misc/postgresql.conf.sample

@@ -157,7 +157,7 @@
 #wal_sync_method = fsync		# the default is the first option
 					# supported by the operating system:
 					#   open_datasync
-					#   fdatasync
+					#   fdatasync (default on Linux)
 					#   fsync
 					#   fsync_writethrough
 					#   open_sync

src/include/access/xlogdefs.h

@@ -123,12 +123,12 @@ typedef uint32 TimeLineID;
 #endif
 #endif
 
-#if defined(OPEN_DATASYNC_FLAG)
+#if defined(PLATFORM_DEFAULT_SYNC_METHOD)
+#define DEFAULT_SYNC_METHOD		PLATFORM_DEFAULT_SYNC_METHOD
+#elif defined(OPEN_DATASYNC_FLAG)
 #define DEFAULT_SYNC_METHOD		SYNC_METHOD_OPEN_DSYNC
 #elif defined(HAVE_FDATASYNC)
 #define DEFAULT_SYNC_METHOD		SYNC_METHOD_FDATASYNC
-#elif defined(HAVE_FSYNC_WRITETHROUGH_ONLY)
-#define DEFAULT_SYNC_METHOD		SYNC_METHOD_FSYNC_WRITETHROUGH
 #else
 #define DEFAULT_SYNC_METHOD		SYNC_METHOD_FSYNC
 #endif

src/include/port/linux.h

@@ -12,3 +12,11 @@
  * to have a kernel version test here.
  */
 #define HAVE_LINUX_EIDRM_BUG
+
+/*
+ * Set the default wal_sync_method to fdatasync.  With recent Linux versions,
+ * xlogdefs.h's normal rules will prefer open_datasync, which (a) doesn't
+ * perform better and (b) causes outright failures on ext4 data=journal
+ * filesystems, because those don't support O_DIRECT.
+ */
+#define PLATFORM_DEFAULT_SYNC_METHOD	SYNC_METHOD_FDATASYNC

src/include/port/win32.h

@@ -34,14 +34,18 @@
 /* Must be here to avoid conflicting with prototype in windows.h */
 #define mkdir(a,b)	mkdir(a)
 
-#define HAVE_FSYNC_WRITETHROUGH
-#define HAVE_FSYNC_WRITETHROUGH_ONLY
 #define ftruncate(a,b)	chsize(a,b)
+
+/* Windows doesn't have fsync() as such, use _commit() */
+#define fsync(fd) _commit(fd)
+
 /*
- *	Even though we don't support 'fsync' as a wal_sync_method,
- *	we do fsync() a few other places where _commit() is just fine.
+ * For historical reasons, we allow setting wal_sync_method to
+ * fsync_writethrough on Windows, even though it's really identical to fsync
+ * (both code paths wind up at _commit()).
  */
-#define fsync(fd) _commit(fd)
+#define HAVE_FSYNC_WRITETHROUGH
+#define FSYNC_WRITETHROUGH_IS_FSYNC
 
 #define USES_WINSOCK