[Issue #159] pgdata_bytes for FULL and DELTA remote backups was calculated incorrectly

Author: gsmolk

src/backup.c

@@ -155,6 +155,7 @@ do_backup_instance(PGconn *backup_conn, PGNodeInfo *nodeInfo)
 
 	/* for fancy reporting */
 	time_t		start_time, end_time;
+	char		pretty_bytes[20];
 
 	elog(LOG, "Database backup start");
 	if(current.external_dir_str)
@@ -330,6 +331,20 @@ do_backup_instance(PGconn *backup_conn, PGNodeInfo *nodeInfo)
 		elog(ERROR, "PGDATA is almost empty. Either it was concurrently deleted or "
 			"pg_probackup do not possess sufficient permissions to list PGDATA content");
 
+	/* Calculate pgdata_bytes */
+	for (i = 0; i < parray_num(backup_files_list); i++)
+	{
+		pgFile	   *file = (pgFile *) parray_get(backup_files_list, i);
+
+		if (file->external_dir_num != 0)
+			continue;
+
+		current.pgdata_bytes += file->size;
+	}
+
+	pretty_size(current.pgdata_bytes, pretty_bytes, lengthof(pretty_bytes));
+	elog(INFO, "PGDATA size: %s", pretty_bytes);
+
 	/*
 	 * Sort pathname ascending. It is necessary to create intermediate
 	 * directories sequentially.
@@ -547,22 +562,6 @@ do_backup_instance(PGconn *backup_conn, PGNodeInfo *nodeInfo)
 	/* close ssh session in main thread */
 	fio_disconnect();
 
-	/* Calculate pgdata_bytes */
-	for (i = 0; i < parray_num(backup_files_list); i++)
-	{
-		pgFile	   *file = (pgFile *) parray_get(backup_files_list, i);
-
-		/* In case of FULL or DELTA backup we can trust read_size.
-		 * In case of PAGE or PTRACK we are forced to trust datafile size,
-		 * taken at the start of backup.
-		 */
-		if (current.backup_mode == BACKUP_MODE_FULL ||
-			current.backup_mode == BACKUP_MODE_DIFF_DELTA)
-			current.pgdata_bytes += file->read_size;
-		else
-			current.pgdata_bytes += file->size;
-	}
-
 	/* Add archived xlog files into the list of files of this backup */
 	if (stream_wal)
 	{
@@ -699,9 +698,9 @@ int
 do_backup(time_t start_time, bool no_validate,
 			pgSetBackupParams *set_backup_params)
 {
-	PGconn *backup_conn = NULL;
-	PGNodeInfo nodeInfo;
-	char  pretty_data_bytes[20];
+	PGconn		*backup_conn = NULL;
+	PGNodeInfo	nodeInfo;
+	char		pretty_bytes[20];
 
 	/* Initialize PGInfonode */
 	pgNodeInit(&nodeInfo);
@@ -836,10 +835,10 @@ do_backup(time_t start_time, bool no_validate,
 
 	/* Notify user about backup size */
 	if (current.stream)
-		pretty_size(current.data_bytes + current.wal_bytes, pretty_data_bytes, lengthof(pretty_data_bytes));
+		pretty_size(current.data_bytes + current.wal_bytes, pretty_bytes, lengthof(pretty_bytes));
 	else
-		pretty_size(current.data_bytes, pretty_data_bytes, lengthof(pretty_data_bytes));
-	elog(INFO, "Backup %s resident size: %s", base36enc(current.start_time), pretty_data_bytes);
+		pretty_size(current.data_bytes, pretty_bytes, lengthof(pretty_bytes));
+	elog(INFO, "Backup %s resident size: %s", base36enc(current.start_time), pretty_bytes);
 
 	if (current.status == BACKUP_STATUS_OK ||
 		current.status == BACKUP_STATUS_DONE)

src/data.c

@@ -453,7 +453,7 @@ compress_and_backup_page(pgFile *file, BlockNumber blknum,
 	char		write_buffer[BLCKSZ+sizeof(header)];
 	char		compressed_page[BLCKSZ*2]; /* compressed page may require more space than uncompressed */
 
-	if(page_state == SkipCurrentPage)
+	if (page_state == SkipCurrentPage)
 		return;
 
 	header.block = blknum;
@@ -641,6 +641,7 @@ backup_data_file(backup_files_arg* arguments,
 					 file->path, rc == PAGE_CHECKSUM_MISMATCH ? "data file checksum mismatch" : strerror(-rc));
 			n_blocks_read = rc;
 
+			file->read_size = n_blocks_read * BLCKSZ;
 			file->uncompressed_size = (n_blocks_read - n_blocks_skipped)*BLCKSZ;
 		}
 		else
@@ -658,6 +659,8 @@ backup_data_file(backup_files_arg* arguments,
 				n_blocks_read++;
 				if (page_state == PageIsTruncated)
 					break;
+
+				file->read_size += BLCKSZ;
 			}
 		}
 		if (backup_mode == BACKUP_MODE_DIFF_DELTA)

src/utils/file.c

@@ -1236,7 +1236,6 @@ int fio_send_pages(FILE* in, FILE* out, pgFile *file,
 			blknum += 1;
 			break;
 		}
-		file->read_size += BLCKSZ;
 	}
 	*nBlocksSkipped = blknum - n_blocks_read;
 	return blknum;