Tolerate timeline switches while "pg_basebackup -X fetch" is running.

If you take a base backup from a standby server with "pg_basebackup -X
fetch", and the timeline switches while the backup is being taken, the
backup used to fail with an error "requested WAL segment %s has already
been removed". This is because the server-side code that sends over the
required WAL files would not construct the WAL filename with the correct
timeline after a switch.

Fix that by using readdir() to scan pg_xlog for all the WAL segments in the
range, regardless of timeline.

Also, include all timeline history files in the backup, if taken with
"-X fetch". That fixes another related bug: If a timeline switch happened
just before the backup was initiated in a standby, the WAL segment
containing the initial checkpoint record contains WAL from the older
timeline too. Recovery will not accept that without a timeline history file
that lists the older timeline.

Backpatch to 9.2. Versions prior to that were not affected as you could not
take a base backup from a standby before 9.2.

src/backend/access/transam/xlog.c

@@ -2797,18 +2797,33 @@ PreallocXlogFiles(XLogRecPtr endptr)
 }
 
 /*
- * Get the segno of the latest removed or recycled WAL segment.
+ * Throws an error if the given log segment has already been removed or
- * Returns 0/0 if no WAL segments have been removed since startup.
+ * recycled. The caller should only pass a segment that it knows to have
+ * existed while the server has been running, as this function always
+ * succeeds if no WAL segments have been removed since startup.
+ * 'tli' is only used in the error message.
  */
 void
-XLogGetLastRemoved(XLogSegNo *segno)
+CheckXLogRemoved(XLogSegNo segno, TimeLineID tli)
 {
 	/* use volatile pointer to prevent code rearrangement */
 	volatile XLogCtlData *xlogctl = XLogCtl;
+	XLogSegNo	lastRemovedSegNo;
 
 	SpinLockAcquire(&xlogctl->info_lck);
-	*segno = xlogctl->lastRemovedSegNo;
+	lastRemovedSegNo = xlogctl->lastRemovedSegNo;
 	SpinLockRelease(&xlogctl->info_lck);
+
+	if (segno <= lastRemovedSegNo)
+	{
+		char		filename[MAXFNAMELEN];
+
+		XLogFileName(filename, tli, segno);
+		ereport(ERROR,
+				(errcode_for_file_access(),
+				 errmsg("requested WAL segment %s has already been removed",
+						filename)));
+	}
 }
 
 /*

src/backend/replication/basebackup.c

@@ -56,14 +56,13 @@ static void base_backup_cleanup(int code, Datum arg);
 static void perform_base_backup(basebackup_options *opt, DIR *tblspcdir);
 static void parse_basebackup_options(List *options, basebackup_options *opt);
 static void SendXlogRecPtrResult(XLogRecPtr ptr);
+static int compareWalFileNames(const void *a, const void *b);
 
 /* Was the backup currently in-progress initiated in recovery mode? */
 static bool backup_started_in_recovery = false;
 
 /*
  * Size of each block sent into the tar stream for larger files.
- *
- * XLogSegSize *MUST* be evenly dividable by this
  */
 #define TAR_SEND_SIZE 32768
 
@@ -227,64 +226,201 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 		 * We've left the last tar file "open", so we can now append the
 		 * required WAL files to it.
 		 */
-		XLogSegNo	logsegno;
+		char		pathbuf[MAXPGPATH];
-		XLogSegNo	endlogsegno;
+		XLogSegNo	segno;
+		XLogSegNo	startsegno;
+		XLogSegNo	endsegno;
 		struct stat statbuf;
-
+		List	   *historyFileList = NIL;
-		MemSet(&statbuf, 0, sizeof(statbuf));
+		List	   *walFileList = NIL;
-		statbuf.st_mode = S_IRUSR | S_IWUSR;
+		char	  **walFiles;
-#ifndef WIN32
+		int			nWalFiles;
-		statbuf.st_uid = geteuid();
+		char		firstoff[MAXFNAMELEN];
-		statbuf.st_gid = getegid();
+		char		lastoff[MAXFNAMELEN];
-#endif
+		DIR		   *dir;
-		statbuf.st_size = XLogSegSize;
+		struct dirent *de;
-		statbuf.st_mtime = time(NULL);
-
-		XLByteToSeg(startptr, logsegno);
-		XLByteToPrevSeg(endptr, endlogsegno);
-
-		while (true)
-		{
-			/* Send another xlog segment */
-			char		fn[MAXPGPATH];
 		int			i;
-
+		ListCell   *lc;
-			XLogFilePath(fn, ThisTimeLineID, logsegno);
+		TimeLineID	tli;
-			_tarWriteHeader(fn, NULL, &statbuf);
-
-			/* Send the actual WAL file contents, block-by-block */
-			for (i = 0; i < XLogSegSize / TAR_SEND_SIZE; i++)
-			{
-				char		buf[TAR_SEND_SIZE];
-				XLogRecPtr	ptr;
-
-				XLogSegNoOffsetToRecPtr(logsegno, TAR_SEND_SIZE * i, ptr);
 
 		/*
-				 * Some old compilers, e.g. gcc 2.95.3/x86, think that passing
+		 * I'd rather not worry about timelines here, so scan pg_xlog and
-				 * a struct in the same function as a longjump might clobber a
+		 * include all WAL files in the range between 'startptr' and 'endptr',
-				 * variable.  bjm 2011-02-04
+		 * regardless of the timeline the file is stamped with. If there are
-				 * http://lists.apple.com/archives/xcode-users/2003/Dec//msg000
+		 * some spurious WAL files belonging to timelines that don't belong
-				 * 51.html
+		 * in this server's history, they will be included too. Normally there
+		 * shouldn't be such files, but if there are, there's little harm in
+		 * including them.
 		 */
-				XLogRead(buf, ThisTimeLineID, ptr, TAR_SEND_SIZE);
+		XLByteToSeg(startptr, startsegno);
-				if (pq_putmessage('d', buf, TAR_SEND_SIZE))
+		XLogFileName(firstoff, ThisTimeLineID, startsegno);
+		XLByteToPrevSeg(endptr, endsegno);
+		XLogFileName(lastoff, ThisTimeLineID, endsegno);
+
+		dir = AllocateDir("pg_xlog");
+		if (!dir)
+			ereport(ERROR,
+					(errmsg("could not open directory \"%s\": %m", "pg_xlog")));
+		while ((de = ReadDir(dir, "pg_xlog")) != NULL)
+		{
+			/* Does it look like a WAL segment, and is it in the range? */
+			if (strlen(de->d_name) == 24 &&
+				strspn(de->d_name, "0123456789ABCDEF") == 24 &&
+				strcmp(de->d_name + 8, firstoff + 8) >= 0 &&
+				strcmp(de->d_name + 8, lastoff + 8) <= 0)
+			{
+				walFileList = lappend(walFileList, pstrdup(de->d_name));
+			}
+			/* Does it look like a timeline history file? */
+			else if (strlen(de->d_name) == 8 + strlen(".history") &&
+					 strspn(de->d_name, "0123456789ABCDEF") == 8 &&
+					 strcmp(de->d_name + 8, ".history") == 0)
+			{
+				historyFileList = lappend(historyFileList, pstrdup(de->d_name));
+			}
+		}
+		FreeDir(dir);
+
+		/*
+		 * Before we go any further, check that none of the WAL segments we
+		 * need were removed.
+		 */
+		CheckXLogRemoved(startsegno, ThisTimeLineID);
+
+		/*
+		 * Put the WAL filenames into an array, and sort. We send the files
+		 * in order from oldest to newest, to reduce the chance that a file
+		 * is recycled before we get a chance to send it over.
+		 */
+		nWalFiles = list_length(walFileList);
+		walFiles = palloc(nWalFiles * sizeof(char *));
+		i = 0;
+		foreach(lc, walFileList)
+		{
+			walFiles[i++] = lfirst(lc);
+		}
+		qsort(walFiles, nWalFiles, sizeof(char *), compareWalFileNames);
+
+		/*
+		 * Sanity check: the first and last segment should cover startptr and
+		 * endptr, with no gaps in between.
+		 */
+		XLogFromFileName(walFiles[0], &tli, &segno);
+		if (segno != startsegno)
+		{
+			char startfname[MAXFNAMELEN];
+			XLogFileName(startfname, ThisTimeLineID, startsegno);
+			ereport(ERROR,
+					(errmsg("could not find WAL file %s", startfname)));
+		}
+		for (i = 0; i < nWalFiles; i++)
+		{
+			XLogSegNo currsegno = segno;
+			XLogSegNo nextsegno = segno + 1;
+
+			XLogFromFileName(walFiles[i], &tli, &segno);
+			if (!(nextsegno == segno || currsegno == segno))
+			{
+				char nextfname[MAXFNAMELEN];
+				XLogFileName(nextfname, ThisTimeLineID, nextsegno);
+				ereport(ERROR,
+						(errmsg("could not find WAL file %s", nextfname)));
+			}
+		}
+		if (segno != endsegno)
+		{
+			char endfname[MAXFNAMELEN];
+			XLogFileName(endfname, ThisTimeLineID, endsegno);
+			ereport(ERROR,
+					(errmsg("could not find WAL file %s", endfname)));
+		}
+
+		/* Ok, we have everything we need. Send the WAL files. */
+		for (i = 0; i < nWalFiles; i++)
+		{
+			FILE	   *fp;
+			char		buf[TAR_SEND_SIZE];
+			size_t		cnt;
+			pgoff_t		len = 0;
+
+			snprintf(pathbuf, MAXPGPATH, XLOGDIR "/%s", walFiles[i]);
+			XLogFromFileName(walFiles[i], &tli, &segno);
+
+			fp = AllocateFile(pathbuf, "rb");
+			if (fp == NULL)
+			{
+				/*
+				 * Most likely reason for this is that the file was already
+				 * removed by a checkpoint, so check for that to get a better
+				 * error message.
+				 */
+				CheckXLogRemoved(segno, tli);
+
+				ereport(ERROR,
+						(errcode_for_file_access(),
+						 errmsg("could not open file \"%s\": %m", pathbuf)));
+			}
+
+			if (fstat(fileno(fp), &statbuf) != 0)
+				ereport(ERROR,
+						(errcode_for_file_access(),
+						 errmsg("could not stat file \"%s\": %m",
+								pathbuf)));
+			if (statbuf.st_size != XLogSegSize)
+			{
+				CheckXLogRemoved(segno, tli);
+				ereport(ERROR,
+						(errcode_for_file_access(),
+						 errmsg("unexpected WAL file size \"%s\"", walFiles[i])));
+			}
+
+			_tarWriteHeader(pathbuf, NULL, &statbuf);
+
+			while ((cnt = fread(buf, 1, Min(sizeof(buf), XLogSegSize - len), fp)) > 0)
+			{
+				CheckXLogRemoved(segno, tli);
+				/* Send the chunk as a CopyData message */
+				if (pq_putmessage('d', buf, cnt))
 					ereport(ERROR,
 							(errmsg("base backup could not send data, aborting backup")));
+
+				len += cnt;
+				if (len == XLogSegSize)
+					break;
+			}
+
+			if (len != XLogSegSize)
+			{
+				CheckXLogRemoved(segno, tli);
+				ereport(ERROR,
+						(errcode_for_file_access(),
+						 errmsg("unexpected WAL file size \"%s\"", walFiles[i])));
+			}
+
+			/* XLogSegSize is a multiple of 512, so no need for padding */
+			FreeFile(fp);
 		}
 
 		/*
-			 * Files are always fixed size, and always end on a 512 byte
+		 * Send timeline history files too. Only the latest timeline history
-			 * boundary, so padding is never necessary.
+		 * file is required for recovery, and even that only if there happens
+		 * to be a timeline switch in the first WAL segment that contains the
+		 * checkpoint record, or if we're taking a base backup from a standby
+		 * server and the target timeline changes while the backup is taken. 
+		 * But they are small and highly useful for debugging purposes, so
+		 * better include them all, always.
 		 */
+		foreach(lc, historyFileList)
+		{
+			char *fname = lfirst(lc);
+			snprintf(pathbuf, MAXPGPATH, XLOGDIR "/%s", fname);
 
+			if (lstat(pathbuf, &statbuf) != 0)
+				ereport(ERROR,
+						(errcode_for_file_access(),
+						 errmsg("could not stat file \"%s\": %m", pathbuf)));
 
-			/* Advance to the next WAL file */
+			sendFile(pathbuf, pathbuf, &statbuf, false);
-			logsegno++;
-
-			/* Have we reached our stop position yet? */
-			if (logsegno > endlogsegno)
-				break;
 		}
 
 		/* Send CopyDone message for the last tar file */
@@ -293,6 +429,19 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
 	SendXlogRecPtrResult(endptr);
 }
 
+/*
+ * qsort comparison function, to compare log/seg portion of WAL segment
+ * filenames, ignoring the timeline portion.
+ */
+static int
+compareWalFileNames(const void *a, const void *b)
+{
+	char *fna = *((char **) a);
+	char *fnb = *((char **) b);
+
+	return strcmp(fna + 8, fnb + 8);
+}
+
 /*
  * Parse the base backup options passed down by the parser
  */

src/backend/replication/walsender.c

@@ -1170,7 +1170,6 @@ XLogRead(char *buf, TimeLineID tli, XLogRecPtr startptr, Size count)
 	char	   *p;
 	XLogRecPtr	recptr;
 	Size		nbytes;
-	XLogSegNo	lastRemovedSegNo;
 	XLogSegNo	segno;
 
 retry:
@@ -1263,13 +1262,8 @@ retry:
 	 * read() succeeds in that case, but the data we tried to read might
 	 * already have been overwritten with new WAL records.
 	 */
-	XLogGetLastRemoved(&lastRemovedSegNo);
 	XLByteToSeg(startptr, segno);
-	if (segno <= lastRemovedSegNo)
+	CheckXLogRemoved(segno, ThisTimeLineID);
-		ereport(ERROR,
-				(errcode_for_file_access(),
-				 errmsg("requested WAL segment %s has already been removed",
-						XLogFileNameP(sendTimeLine, segno))));
 
 	/*
 	 * During recovery, the currently-open WAL file might be replaced with the

src/include/access/xlog.h

@@ -267,7 +267,7 @@ extern bool XLogNeedsFlush(XLogRecPtr RecPtr);
 extern int XLogFileInit(XLogSegNo segno, bool *use_existent, bool use_lock);
 extern int	XLogFileOpen(XLogSegNo segno);
 
-extern void XLogGetLastRemoved(XLogSegNo *segno);
+extern void CheckXLogRemoved(XLogSegNo segno, TimeLineID tli);
 extern void XLogSetAsyncXactLSN(XLogRecPtr record);
 
 extern Buffer RestoreBackupBlock(XLogRecPtr lsn, XLogRecord *record,