Check the file system on postmaster startup and report any unreferenced

files in the server log.

Heikki Linnakangas

doc/src/sgml/maintenance.sgml

@@ -1,5 +1,5 @@
 <!--
-$PostgreSQL: pgsql/doc/src/sgml/maintenance.sgml,v 1.41 2005/02/20 02:21:26 tgl Exp $
+$PostgreSQL: pgsql/doc/src/sgml/maintenance.sgml,v 1.42 2005/05/02 18:26:52 momjian Exp $
 -->
 
 <chapter id="maintenance">
@@ -474,6 +474,23 @@ HINT:  Stop the postmaster and use a standalone backend to VACUUM in "mydb".
   </para>
  </sect1>
 
+ <sect1 id="check-files-after-crash">
+  <title>Check files after crash</title>
+
+  <indexterm zone="check-files-after-crash">
+   <primary>stale file</primary>
+  </indexterm>
+
+  <para>
+   <productname>PostgreSQL</productname> recovers automatically after crash
+   using the write-ahead log (see <xref linkend="wal">) and no manual 
+   operations are normally needed. However, if there was a transaction running 
+   when the crash occured that created or dropped a relation, the 
+   transaction might have left a stale file in the data directory. If this 
+   happens, you will get a notice in the log file stating which files can be 
+   deleted.
+  </para>
+ </sect1>
 
  <sect1 id="logfile-maintenance">
   <title>Log File Maintenance</title>

src/backend/access/transam/xlog.c

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2005, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/backend/access/transam/xlog.c,v 1.189 2005/04/28 21:47:10 tgl Exp $
+ * $PostgreSQL: pgsql/src/backend/access/transam/xlog.c,v 1.190 2005/05/02 18:26:52 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -43,6 +43,7 @@
 #include "utils/builtins.h"
 #include "utils/guc.h"
 #include "utils/relcache.h"
+#include "utils/flatfiles.h"
 
 
 /*
@@ -4525,6 +4526,8 @@ StartupXLOG(void)
 
 		CreateCheckPoint(true, true);
 
+		CheckStaleRelFiles();
+
 		/*
 		 * Close down recovery environment
 		 */
@@ -4536,6 +4539,12 @@ StartupXLOG(void)
 		 */
 		remove_backup_label();
 	}
+	else
+	{
+		XLogInitRelationCache();
+		CheckStaleRelFiles();
+		XLogCloseRelationCache();
+	}
 
 	/*
 	 * Preallocate additional log files, if wanted.

src/backend/catalog/catalog.c

@@ -9,7 +9,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/catalog/catalog.c,v 1.59 2005/04/14 20:03:23 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/catalog/catalog.c,v 1.60 2005/05/02 18:26:53 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -106,6 +106,39 @@ GetDatabasePath(Oid dbNode, Oid spcNode)
 	return path;
 }
 
+/*
+ * GetTablespacePath	- construct path to a tablespace symbolic link
+ *
+ * Result is a palloc'd string.
+ *
+ * XXX this must agree with relpath and GetDatabasePath!
+ */
+char *
+GetTablespacePath(Oid spcNode)
+{
+	int			pathlen;
+	char	   *path;
+
+	Assert(spcNode != GLOBALTABLESPACE_OID);
+
+	if (spcNode == DEFAULTTABLESPACE_OID)
+	{
+		/* The default tablespace is {datadir}/base */
+		pathlen = strlen(DataDir) + 5 + 1;
+		path = (char *) palloc(pathlen);
+		snprintf(path, pathlen, "%s/base",
+				 DataDir);
+	}
+	else
+	{
+		/* All other tablespaces have symlinks in pg_tblspc */
+		pathlen = strlen(DataDir) + 11 + OIDCHARS + 1;
+		path = (char *) palloc(pathlen);
+		snprintf(path, pathlen, "%s/pg_tblspc/%u",
+				 DataDir, spcNode);
+	}
+	return path;
+}
 
 /*
  * IsSystemRelation

src/backend/commands/tablespace.c

@@ -37,7 +37,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/commands/tablespace.c,v 1.17 2005/04/14 20:03:24 tgl Exp $
+ *	  $PostgreSQL: pgsql/src/backend/commands/tablespace.c,v 1.18 2005/05/02 18:26:53 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -341,8 +341,7 @@ CreateTableSpace(CreateTableSpaceStmt *stmt)
 	/*
 	 * All seems well, create the symlink
 	 */
-	linkloc = (char *) palloc(strlen(DataDir) + 11 + 10 + 1);
+	linkloc = GetTablespacePath(tablespaceoid);
-	sprintf(linkloc, "%s/pg_tblspc/%u", DataDir, tablespaceoid);
 
 	if (symlink(location, linkloc) < 0)
 		ereport(ERROR,
@@ -495,8 +494,7 @@ remove_tablespace_directories(Oid tablespaceoid, bool redo)
 	char	   *subfile;
 	struct stat st;
 
-	location = (char *) palloc(strlen(DataDir) + 11 + 10 + 1);
+	location = GetTablespacePath(tablespaceoid);
-	sprintf(location, "%s/pg_tblspc/%u", DataDir, tablespaceoid);
 
 	/*
 	 * Check if the tablespace still contains any files.  We try to rmdir
@@ -1036,8 +1034,7 @@ tblspc_redo(XLogRecPtr lsn, XLogRecord *record)
 		set_short_version(location);
 
 		/* Create the symlink if not already present */
-		linkloc = (char *) palloc(strlen(DataDir) + 11 + 10 + 1);
+		linkloc = GetTablespacePath(xlrec->ts_id);
-		sprintf(linkloc, "%s/pg_tblspc/%u", DataDir, xlrec->ts_id);
 
 		if (symlink(location, linkloc) < 0)
 		{

src/backend/utils/adt/misc.c

@@ -8,7 +8,7 @@
  *
  *
  * IDENTIFICATION
- *	  $PostgreSQL: pgsql/src/backend/utils/adt/misc.c,v 1.40 2004/12/31 22:01:22 pgsql Exp $
+ *	  $PostgreSQL: pgsql/src/backend/utils/adt/misc.c,v 1.41 2005/05/02 18:26:53 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -26,6 +26,7 @@
 #include "funcapi.h"
 #include "catalog/pg_type.h"
 #include "catalog/pg_tablespace.h"
+#include "catalog/catalog.h"
 
 #define atooid(x)  ((Oid) strtoul((x), NULL, 10))
 
@@ -144,11 +145,6 @@ pg_tablespace_databases(PG_FUNCTION_ARGS)
 
 		fctx = palloc(sizeof(ts_db_fctx));
 
-		/*
-		 * size = path length + tablespace dirname length + 2 dir sep
-		 * chars + oid + terminator
-		 */
-		fctx->location = (char *) palloc(strlen(DataDir) + 11 + 10 + 1);
 		if (tablespaceOid == GLOBALTABLESPACE_OID)
 		{
 			fctx->dirdesc = NULL;
@@ -157,12 +153,7 @@ pg_tablespace_databases(PG_FUNCTION_ARGS)
 		}
 		else
 		{
-			if (tablespaceOid == DEFAULTTABLESPACE_OID)
+			fctx->location = GetTablespacePath(tablespaceOid);
-				sprintf(fctx->location, "%s/base", DataDir);
-			else
-				sprintf(fctx->location, "%s/pg_tblspc/%u", DataDir,
-						tablespaceOid);
-
 			fctx->dirdesc = AllocateDir(fctx->location);
 
 			if (!fctx->dirdesc)

src/backend/utils/init/Makefile

@@ -4,7 +4,7 @@
 #    Makefile for utils/init
 #
 # IDENTIFICATION
-#    $PostgreSQL: pgsql/src/backend/utils/init/Makefile,v 1.18 2005/02/20 02:22:00 tgl Exp $
+#    $PostgreSQL: pgsql/src/backend/utils/init/Makefile,v 1.19 2005/05/02 18:26:53 momjian Exp $
 #
 #-------------------------------------------------------------------------
 
@@ -12,7 +12,7 @@ subdir = src/backend/utils/init
 top_builddir = ../../../..
 include $(top_builddir)/src/Makefile.global
 
-OBJS = flatfiles.o globals.o miscinit.o postinit.o
+OBJS = flatfiles.o globals.o miscinit.o postinit.o checkfiles.o
 
 all: SUBSYS.o
 

src/backend/utils/init/checkfiles.c

@@ -0,0 +1,225 @@
+/*-------------------------------------------------------------------------
+ *
+ *	checkfiles.c
+ *	  support to clean up stale relation files on crash recovery
+ *
+ *	If a backend crashes while in a transaction that has created or
+ *	deleted a relfilenode, a stale file can be left over in the data
+ *	directory. This file contains routines to clean up those stale
+ *	files on recovery.
+ *
+ *	This adds a 17% increase in startup cost for 100 empty databases.  bjm
+ *	One optimization would be to create a 'dirty' file on a postmaster recovery
+ *	and remove the dirty flag only when a clean startup detects no unreferenced
+ *	files, and use the 'dirty' flag to determine if we should run this on
+ *	a clean startup.
+ *
+ * $PostgreSQL: pgsql/src/backend/utils/init/checkfiles.c,v 1.1 2005/05/02 18:26:53 momjian Exp $
+ *
+ *-------------------------------------------------------------------------
+ */
+#include "postgres.h"
+
+#include "storage/fd.h"
+
+#include "utils/flatfiles.h"
+#include "miscadmin.h"
+#include "catalog/pg_tablespace.h"
+#include "catalog/catalog.h"
+#include "access/skey.h"
+#include "utils/fmgroids.h"
+#include "access/relscan.h"
+#include "access/heapam.h"
+#include "utils/resowner.h"
+
+static void CheckStaleRelFilesFrom(Oid tablespaceoid, Oid dboid);
+static void CheckStaleRelFilesFromTablespace(Oid tablespaceoid);
+
+/* Like AllocateDir, but ereports on failure */
+static DIR *
+AllocateDirChecked(char *path)
+{
+	DIR		   *dirdesc = AllocateDir(path);
+
+	if (dirdesc == NULL)
+		ereport(ERROR,
+				(errcode_for_file_access(),
+				 errmsg("could not open directory \"%s\": %m",
+						path)));
+	return dirdesc;
+}
+
+/*
+ * Scan through all tablespaces for relations left over
+ * by aborted transactions.
+ *
+ * For example, if a transaction issues
+ * BEGIN; CREATE TABLE foobar ();
+ * and then the backend crashes, the file is left in the
+ * tablespace until CheckStaleRelFiles deletes it.
+ */
+void
+CheckStaleRelFiles(void)
+{
+	DIR		   *dirdesc;
+	struct dirent *de;
+	char	   *path;
+	int			pathlen;
+
+	pathlen = strlen(DataDir) + 11 + 1;
+	path = (char *) palloc(pathlen);
+	snprintf(path, pathlen, "%s/pg_tblspc/", DataDir);
+	dirdesc = AllocateDirChecked(path);
+	while ((de = readdir(dirdesc)) != NULL)
+	{
+		char	   *invalid;
+		Oid			tablespaceoid;
+
+		/* Check that the directory name looks like valid tablespace link.	*/
+		tablespaceoid = (Oid) strtol(de->d_name, &invalid, 10);
+		if (invalid[0] == '\0')
+			CheckStaleRelFilesFromTablespace(tablespaceoid);
+	}
+	FreeDir(dirdesc);
+	pfree(path);
+
+	CheckStaleRelFilesFromTablespace(DEFAULTTABLESPACE_OID);
+}
+
+/* Scan a specific tablespace for stale relations */
+static void
+CheckStaleRelFilesFromTablespace(Oid tablespaceoid)
+{
+	DIR		   *dirdesc;
+	struct dirent *de;
+	char	   *path;
+
+	path = GetTablespacePath(tablespaceoid);
+
+	dirdesc = AllocateDirChecked(path);
+	while ((de = readdir(dirdesc)) != NULL)
+	{
+		char	   *invalid;
+		Oid			dboid;
+
+		dboid = (Oid) strtol(de->d_name, &invalid, 10);
+		if (invalid[0] == '\0')
+			CheckStaleRelFilesFrom(tablespaceoid, dboid);
+	}
+	FreeDir(dirdesc);
+	pfree(path);
+}
+
+/* Scan a specific database in a specific tablespace for stale relations.
+ *
+ * First, pg_class for the database is opened, and the relfilenodes of all
+ * relations mentioned there are stored in a hash table.
+ *
+ * Then the directory is scanned. Every file in the directory that's not
+ * found in pg_class (the hash table) is logged.
+ */
+static void
+CheckStaleRelFilesFrom(Oid tablespaceoid, Oid dboid)
+{
+	DIR		   *dirdesc;
+	struct dirent *de;
+	HASHCTL		hashctl;
+	HTAB	   *relfilenodeHash;
+	MemoryContext mcxt;
+	RelFileNode rnode;
+	char	   *path;
+
+	/*
+	 * We create a private memory context so that we can easily deallocate the
+	 * hash table and its contents
+	 */
+	mcxt = AllocSetContextCreate(TopMemoryContext, "CheckStaleRelFiles",
+								 ALLOCSET_DEFAULT_MINSIZE,
+								 ALLOCSET_DEFAULT_INITSIZE,
+								 ALLOCSET_DEFAULT_MAXSIZE);
+
+	hashctl.hash = tag_hash;
+
+	/*
+	 * The entry contents is not used for anything, we just check if an oid is
+	 * in the hash table or not.
+	 */
+	hashctl.keysize = sizeof(Oid);
+	hashctl.entrysize = 1;
+	hashctl.hcxt = mcxt;
+	relfilenodeHash = hash_create("relfilenodeHash", 100, &hashctl,
+								  HASH_FUNCTION
+								  | HASH_ELEM | HASH_CONTEXT);
+
+	/* Read all relfilenodes from pg_class into the hash table */
+	{
+		ResourceOwner owner,
+					oldowner;
+		Relation	rel;
+		HeapScanDesc scan;
+		HeapTuple	tuple;
+
+		/* Need a resowner to keep the heapam and buffer code happy */
+		owner = ResourceOwnerCreate(NULL, "CheckStaleRelFiles");
+		oldowner = CurrentResourceOwner;
+		CurrentResourceOwner = owner;
+
+		rnode.spcNode = tablespaceoid;
+		rnode.dbNode = dboid;
+		rnode.relNode = RelationRelationId;
+		rel = XLogOpenRelation(true, 0, rnode);
+
+		scan = heap_beginscan(rel, SnapshotNow, 0, NULL);
+		while ((tuple = heap_getnext(scan, ForwardScanDirection)) != NULL)
+		{
+			Form_pg_class classform = (Form_pg_class) GETSTRUCT(tuple);
+
+			hash_search(relfilenodeHash, &classform->relfilenode,
+						HASH_ENTER, NULL);
+		}
+		heap_endscan(scan);
+
+		XLogCloseRelation(rnode);
+		CurrentResourceOwner = oldowner;
+		ResourceOwnerDelete(owner);
+	}
+
+	/* Scan the directory */
+	path = GetDatabasePath(dboid, tablespaceoid);
+
+	dirdesc = AllocateDirChecked(path);
+	while ((de = readdir(dirdesc)) != NULL)
+	{
+		char	   *invalid;
+		Oid			relfilenode;
+
+		relfilenode = strtol(de->d_name, &invalid, 10);
+		if (invalid[0] == '\0')
+		{
+			/*
+			 * Filename was a valid number, check if pg_class knows about it
+			 */
+			if (hash_search(relfilenodeHash, &relfilenode,
+							HASH_FIND, NULL) == NULL)
+			{
+				char	   *filepath;
+
+				rnode.spcNode = tablespaceoid;
+				rnode.dbNode = dboid;
+				rnode.relNode = relfilenode;
+
+				filepath = relpath(rnode);
+
+				ereport(LOG,
+						(errcode_for_file_access(),
+						 errmsg("The table or index file \"%s\" is stale and can be safely removed",
+								filepath)));
+				pfree(filepath);
+			}
+		}
+	}
+	FreeDir(dirdesc);
+	pfree(path);
+	hash_destroy(relfilenodeHash);
+	MemoryContextDelete(mcxt);
+}

src/include/catalog/catalog.h

@@ -7,7 +7,7 @@
  * Portions Copyright (c) 1996-2005, PostgreSQL Global Development Group
  * Portions Copyright (c) 1994, Regents of the University of California
  *
- * $PostgreSQL: pgsql/src/include/catalog/catalog.h,v 1.30 2004/12/31 22:03:24 pgsql Exp $
+ * $PostgreSQL: pgsql/src/include/catalog/catalog.h,v 1.31 2005/05/02 18:26:54 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -19,6 +19,7 @@
 
 extern char *relpath(RelFileNode rnode);
 extern char *GetDatabasePath(Oid dbNode, Oid spcNode);
+extern char *GetTablespacePath(Oid spcNode);
 
 extern bool IsSystemRelation(Relation relation);
 extern bool IsToastRelation(Relation relation);

src/include/utils/flatfiles.h

@@ -4,7 +4,7 @@
  *	  Routines for maintaining "flat file" images of the shared catalogs.
  *
  *
- * $PostgreSQL: pgsql/src/include/utils/flatfiles.h,v 1.1 2005/02/20 02:22:07 tgl Exp $
+ * $PostgreSQL: pgsql/src/include/utils/flatfiles.h,v 1.2 2005/05/02 18:26:54 momjian Exp $
  *
  *-------------------------------------------------------------------------
  */
@@ -30,4 +30,7 @@ extern void AtEOSubXact_UpdateFlatFiles(bool isCommit,
 
 extern Datum flatfile_update_trigger(PG_FUNCTION_ARGS);
 
+/* from checkfiles.c */
+extern void CheckStaleRelFiles(void);
+
 #endif   /* FLATFILES_H */