XLOG (and related) changes:

* Store two past checkpoint locations, not just one, in pg_control.
  On startup, we fall back to the older checkpoint if the newer one
  is unreadable.  Also, a physical copy of the newest checkpoint record
  is kept in pg_control for possible use in disaster recovery (ie,
  complete loss of pg_xlog).  Also add a version number for pg_control
  itself.  Remove archdir from pg_control; it ought to be a GUC
  parameter, not a special case (not that it's implemented yet anyway).

* Suppress successive checkpoint records when nothing has been entered
  in the WAL log since the last one.  This is not so much to avoid I/O
  as to make it actually useful to keep track of the last two
  checkpoints.  If the things are right next to each other then there's
  not a lot of redundancy gained...

* Change CRC scheme to a true 64-bit CRC, not a pair of 32-bit CRCs
  on alternate bytes.  Polynomial borrowed from ECMA DLT1 standard.

* Fix XLOG record length handling so that it will work at BLCKSZ = 32k.

* Change XID allocation to work more like OID allocation.  (This is of
  dubious necessity, but I think it's a good idea anyway.)

* Fix a number of minor bugs, such as off-by-one logic for XLOG file
  wraparound at the 4 gig mark.

* Add documentation and clean up some coding infelicities; move file
  format declarations out to include files where planned contrib
  utilities can get at them.

* Checkpoint will now occur every CHECKPOINT_SEGMENTS log segments or
  every CHECKPOINT_TIMEOUT seconds, whichever comes first.  It is also
  possible to force a checkpoint by sending SIGUSR1 to the postmaster
  (undocumented feature...)

* Defend against kill -9 postmaster by storing shmem block's key and ID
  in postmaster.pid lockfile, and checking at startup to ensure that no
  processes are still connected to old shmem block (if it still exists).

* Switch backends to accept SIGQUIT rather than SIGUSR1 for emergency
  stop, for symmetry with postmaster and xlog utilities.  Clean up signal
  handling in bootstrap.c so that xlog utilities launched by postmaster
  will react to signals better.

* Standalone bootstrap now grabs lockfile in target directory, as added
  insurance against running it in parallel with live postmaster.

Author: tglsfdc

doc/src/sgml/ref/checkpoint.sgml

@@ -1,4 +1,4 @@
-<!-- $Header: /cvsroot/pgsql/doc/src/sgml/ref/checkpoint.sgml,v 1.3 2001/01/27 10:19:52 petere Exp $ -->
+<!-- $Header: /cvsroot/pgsql/doc/src/sgml/ref/checkpoint.sgml,v 1.4 2001/03/13 01:17:05 tgl Exp $ -->
 
 <refentry id="sql-checkpoint">
  <docinfo>
@@ -26,11 +26,12 @@ CHECKPOINT
 
   <para>
    Write-Ahead Logging (WAL) puts a checkpoint in the transaction log
-   every 300 seconds by default.  (This may be changed by the run-time
-   configuration option <parameter>CHECKPOINT_TIMEOUT</parameter>.)
-   The <command>CHECKPOINT</command> command forces a checkpoint at
-   the point at which the command is issued.  The next automatic
-   checkpoint will still happen after the original cycle expires.
+   every so often. (To adjust the automatic checkpoint interval, see
+   the run-time
+   configuration options <parameter>CHECKPOINT_SEGMENTS</parameter>
+   and <parameter>CHECKPOINT_TIMEOUT</parameter>.)
+   The <command>CHECKPOINT</command> command forces an immediate checkpoint
+   when the command is issued, without waiting for a scheduled checkpoint.
   </para>
 
   <para>

doc/src/sgml/runtime.sgml

@@ -1,5 +1,5 @@
 <!--
-$Header: /cvsroot/pgsql/doc/src/sgml/runtime.sgml,v 1.55 2001/02/18 05:30:12 tgl Exp $
+$Header: /cvsroot/pgsql/doc/src/sgml/runtime.sgml,v 1.56 2001/03/13 01:17:05 tgl Exp $
 -->
 
 <Chapter Id="runtime">
@@ -976,6 +976,11 @@ env PGOPTIONS='-c geqo=off' psql
 	fsyncs because of performance problems, you may wish to reconsider
 	your choice.
        </para>
+
+       <para>
+	This option can only be set at server start or in the
+	<filename>postgresql.conf</filename> file.
+       </para>
       </listitem>
      </varlistentry>
 
@@ -1192,11 +1197,25 @@ env PGOPTIONS='-c geqo=off' psql
     tuning.
 
     <variablelist>
+     <varlistentry>
+      <term>CHECKPOINT_SEGMENTS (<type>integer</type>)</term>
+      <listitem>
+       <para>
+        Maximum distance between automatic WAL checkpoints, in logfile
+	segments (each segment is normally 16 megabytes).
+	This option can only be set at server start or in the
+	<filename>postgresql.conf</filename> file.
+       </para>
+      </listitem>
+     </varlistentry>
+
      <varlistentry>
       <term>CHECKPOINT_TIMEOUT (<type>integer</type>)</term>
       <listitem>
        <para>
-        Frequency of automatic WAL checkpoints, in seconds.
+        Maximum time between automatic WAL checkpoints, in seconds.
+	This option can only be set at server start or in the
+	<filename>postgresql.conf</filename> file.
        </para>
       </listitem>
      </varlistentry>
@@ -1205,8 +1224,8 @@ env PGOPTIONS='-c geqo=off' psql
       <term>WAL_BUFFERS (<type>integer</type>)</term>
       <listitem>
        <para>
-        Number of buffers for WAL.  This option can only be set at
-        server start.
+        Number of disk-page buffers for WAL log.  This option can only be set
+	at server start.
        </para>
       </listitem>
      </varlistentry>
@@ -1226,7 +1245,8 @@ env PGOPTIONS='-c geqo=off' psql
       <listitem>
        <para>
         Number of log files that are created in advance at checkpoint
-        time.  This option can only be set at server start.
+        time.  This option can only be set at server start or in the
+	<filename>postgresql.conf</filename> file.
        </para>
       </listitem>
      </varlistentry>
@@ -1909,7 +1929,7 @@ default:\
      <listitem>
       <para>
        This is the <firstterm>Immediate Shutdown</firstterm> which
-       will cause the postmaster to send a SIGUSR1 to all backends and
+       will cause the postmaster to send a SIGQUIT to all backends and
        exit immediately (without properly shutting down the database
        system). When WAL is implemented, this will lead to recovery on
        start-up. Right now it's not recommendable to use this option.

doc/src/sgml/wal.sgml

@@ -1,4 +1,4 @@
-<!-- $Header: /cvsroot/pgsql/doc/src/sgml/wal.sgml,v 1.3 2001/02/26 00:50:07 tgl Exp $ -->
+<!-- $Header: /cvsroot/pgsql/doc/src/sgml/wal.sgml,v 1.4 2001/03/13 01:17:05 tgl Exp $ -->
 
 <chapter id="wal">
  <title>Write-Ahead Logging (<acronym>WAL</acronym>)</title>
@@ -257,7 +257,7 @@
    The <acronym>WAL</acronym> log is held on the disk as a set of 16
    MB files called <firstterm>segments</firstterm>.  By default a new
    segment is created only if more than 75% of the current segment is
-   used. One can instruct the server to create up to 64 log segments
+   used. One can instruct the server to pre-create up to 64 log segments
    at checkpoint time by modifying the <varname>WAL_FILES</varname>
    configuration parameter.
   </para>
@@ -272,11 +272,12 @@
   </para>
 
   <para>
-   By default, the postmaster spawns a special backend process to
-   create the next checkpoint 300 seconds after the previous
-   checkpoint's creation.  One can change this interval by modifying
-   the <varname>CHECKPOINT_TIMEOUT</varname> parameter.  It is also
-   possible to force a checkpoint by using the SQL command
+   The postmaster spawns a special backend process every so often
+   to create the next checkpoint.  A checkpoint is created every
+   <varname>CHECKPOINT_SEGMENTS</varname> log segments, or every
+   <varname>CHECKPOINT_TIMEOUT</varname> seconds, whichever comes first.
+   The default settings are 3 segments and 300 seconds respectively.
+   It is also possible to force a checkpoint by using the SQL command
    <command>CHECKPOINT</command>.
   </para>
 

src/backend/access/transam/transam.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/access/transam/transam.c,v 1.39 2001/01/24 19:42:51 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/access/transam/transam.c,v 1.40 2001/03/13 01:17:05 tgl Exp $
  *
  * NOTES
  *	  This file contains the high level access-method interface to the
@@ -430,6 +430,7 @@ InitializeTransactionLog(void)
 		Assert(!IsUnderPostmaster && 
 				ShmemVariableCache->nextXid <= FirstTransactionId);
 		ShmemVariableCache->nextXid = FirstTransactionId;
+		ShmemVariableCache->xidCount = 0; /* force an XLOG rec right away */
 	}
 	else if (RecoveryCheckingEnabled())
 	{

src/backend/access/transam/varsup.c

@@ -6,20 +6,25 @@
  * Copyright (c) 2000, PostgreSQL Global Development Group
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/access/transam/varsup.c,v 1.35 2001/01/24 19:42:51 momjian Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/access/transam/varsup.c,v 1.36 2001/03/13 01:17:05 tgl Exp $
  *
  *-------------------------------------------------------------------------
  */
 
 #include "postgres.h"
 
 #include "access/transam.h"
+#include "access/xlog.h"
 #include "storage/proc.h"
 
-SPINLOCK OidGenLockId;
 
-extern SPINLOCK XidGenLockId;
-extern void XLogPutNextOid(Oid nextOid);
+/* Number of XIDs and OIDs to prefetch (preallocate) per XLOG write */
+#define VAR_XID_PREFETCH		1024
+#define VAR_OID_PREFETCH		8192
+
+/* Spinlocks for serializing generation of XIDs and OIDs, respectively */
+SPINLOCK XidGenLockId;
+SPINLOCK OidGenLockId;
 
 /* pointer to "variable cache" in shared memory (set up by shmem.c) */
 VariableCache ShmemVariableCache = NULL;
@@ -38,23 +43,31 @@ GetNewTransactionId(TransactionId *xid)
 	}
 
 	SpinAcquire(XidGenLockId);
+
+	/* If we run out of logged for use xids then we must log more */
+	if (ShmemVariableCache->xidCount == 0)
+	{
+		XLogPutNextXid(ShmemVariableCache->nextXid + VAR_XID_PREFETCH);
+		ShmemVariableCache->xidCount = VAR_XID_PREFETCH;
+	}
+
 	*xid = ShmemVariableCache->nextXid;
-	(ShmemVariableCache->nextXid)++;
 
-	if (MyProc != (PROC *) NULL)
-		MyProc->xid = *xid;
+	(ShmemVariableCache->nextXid)++;
+	(ShmemVariableCache->xidCount)--;
 
 	SpinRelease(XidGenLockId);
 
+	if (MyProc != (PROC *) NULL)
+		MyProc->xid = *xid;
 }
 
 /*
- * Like GetNewTransactionId reads nextXid but don't fetch it.
+ * Read nextXid but don't allocate it.
  */
 void
 ReadNewTransactionId(TransactionId *xid)
 {
-
 	/*
 	 * During bootstrap initialization, we return the special
 	 * bootstrap transaction id.
@@ -68,23 +81,21 @@ ReadNewTransactionId(TransactionId *xid)
 	SpinAcquire(XidGenLockId);
 	*xid = ShmemVariableCache->nextXid;
 	SpinRelease(XidGenLockId);
-
 }
 
 /* ----------------------------------------------------------------
  *					object id generation support
  * ----------------------------------------------------------------
  */
 
-#define VAR_OID_PREFETCH		8192
 static Oid lastSeenOid = InvalidOid;
 
 void
 GetNewObjectId(Oid *oid_return)
 {
 	SpinAcquire(OidGenLockId);
 
-	/* If we run out of logged for use oids then we log more */
+	/* If we run out of logged for use oids then we must log more */
 	if (ShmemVariableCache->oidCount == 0)
 	{
 		XLogPutNextOid(ShmemVariableCache->nextOid + VAR_OID_PREFETCH);
@@ -103,11 +114,11 @@ GetNewObjectId(Oid *oid_return)
 void
 CheckMaxObjectId(Oid assigned_oid)
 {
-
 	if (lastSeenOid != InvalidOid && assigned_oid < lastSeenOid)
 		return;
 
 	SpinAcquire(OidGenLockId);
+
 	if (assigned_oid < ShmemVariableCache->nextOid)
 	{
 		lastSeenOid = ShmemVariableCache->nextOid - 1;
@@ -138,5 +149,4 @@ CheckMaxObjectId(Oid assigned_oid)
 	ShmemVariableCache->nextOid = assigned_oid + 1;
 
 	SpinRelease(OidGenLockId);
-
 }

src/backend/access/transam/xact.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $Header: /cvsroot/pgsql/src/backend/access/transam/xact.c,v 1.98 2001/02/26 00:50:07 tgl Exp $
+ *	  $Header: /cvsroot/pgsql/src/backend/access/transam/xact.c,v 1.99 2001/03/13 01:17:05 tgl Exp $
  *
  * NOTES
  *		Transaction aborts can now occur two ways:
@@ -706,14 +706,18 @@ RecordTransactionCommit()
 		}
 
 		XLogFlush(recptr);
+
+		/* Break the chain of back-links in the XLOG records I output */
 		MyLastRecPtr.xrecoff = 0;
 
 		TransactionIdCommit(xid);
 
-		MyProc->logRec.xrecoff = 0;
 		END_CRIT_SECTION();
 	}
 
+	/* Show myself as out of the transaction in PROC array */
+	MyProc->logRec.xrecoff = 0;
+
 	if (leak)
 		ResetBufferPool(true);
 }
@@ -802,6 +806,10 @@ RecordTransactionAbort(void)
 {
 	TransactionId xid = GetCurrentTransactionId();
 
+	/*
+	 * Double check here is to catch case that we aborted partway through
+	 * RecordTransactionCommit ...
+	 */
 	if (MyLastRecPtr.xrecoff != 0 && !TransactionIdDidCommit(xid))
 	{
 		XLogRecData		rdata;
@@ -815,13 +823,19 @@ RecordTransactionAbort(void)
 		rdata.next = NULL;
 
 		START_CRIT_SECTION();
+
 		recptr = XLogInsert(RM_XACT_ID, XLOG_XACT_ABORT, &rdata);
 
 		TransactionIdAbort(xid);
-		MyProc->logRec.xrecoff = 0;
+
 		END_CRIT_SECTION();
 	}
 
+	/* Break the chain of back-links in the XLOG records I output */
+	MyLastRecPtr.xrecoff = 0;
+	/* Show myself as out of the transaction in PROC array */
+	MyProc->logRec.xrecoff = 0;
+
 	/*
 	 * Tell bufmgr and smgr to release resources.
 	 */
@@ -1187,10 +1201,6 @@ AbortTransaction(void)
 	AtEOXact_CatCache(false);
 	AtAbort_Memory();
 	AtEOXact_Files();
-
-	/* Here we'll rollback xaction changes */
-	MyLastRecPtr.xrecoff = 0;
-
 	AtAbort_Locks();
 
 	SharedBufferChanged = false; /* safest place to do it */