[Issue #228] fix merge

Author: gsmolk

src/backup.c

@@ -2161,7 +2161,7 @@ backup_files(void *arg)
 								 arguments->nodeInfo->checksum_version,
 								 arguments->nodeInfo->ptrack_version_num,
 								 arguments->nodeInfo->ptrack_schema,
-								 arguments->hdr_map, true);
+								 arguments->hdr_map, false);
 		}
 		else
 		{

src/catalog.c

@@ -474,9 +474,11 @@ catalog_get_backup_list(const char *instance_name, time_t requested_backup_id)
 		backup->database_dir = pgut_malloc(MAXPGPATH);
 		join_path_components(backup->database_dir, backup->root_dir, DATABASE_DIR);
 
-		/* block header map */
+		/* block header map, TODO: move to separate function */
 		backup->hdr_map.path = pgut_malloc(MAXPGPATH);
 		join_path_components(backup->hdr_map.path, backup->database_dir, HEADER_MAP);
+		backup->hdr_map.path_tmp = pgut_malloc(MAXPGPATH);
+		join_path_components(backup->hdr_map.path_tmp, backup->database_dir, HEADER_MAP_TMP);
 		backup->hdr_map.fp = NULL;
 
 		/* TODO: save encoded backup id */
@@ -855,6 +857,8 @@ pgBackupCreateDir(pgBackup *backup)
 	/* block header map */
 	backup->hdr_map.path = pgut_malloc(MAXPGPATH);
 	join_path_components(backup->hdr_map.path, backup->database_dir, HEADER_MAP);
+	backup->hdr_map.path_tmp = pgut_malloc(MAXPGPATH);
+	join_path_components(backup->hdr_map.path_tmp, backup->database_dir, HEADER_MAP_TMP);
 	backup->hdr_map.fp = NULL;
 
 	/* create directories for actual backup files */
@@ -2282,6 +2286,7 @@ pgBackupInit(pgBackup *backup)
 	backup->content_crc = 0;
 
 	backup->hdr_map.path = NULL;
+	backup->hdr_map.path_tmp = NULL;
 	backup->hdr_map.fp = NULL;
 	backup->hdr_map.mutex = (pthread_mutex_t)PTHREAD_MUTEX_INITIALIZER;
 }
@@ -2298,6 +2303,7 @@ pgBackupFree(void *backup)
 	pg_free(b->database_dir);
 	pg_free(b->note);
 	pg_free(b->hdr_map.path);
+	pg_free(b->hdr_map.path_tmp);
 	pg_free(backup);
 }
 

src/data.c

@@ -31,8 +31,6 @@ typedef struct DataPage
 	char		data[BLCKSZ];
 } DataPage;
 
-static BackupPageHeader2* get_data_file_headers(HeaderMap *hdr_map, pgFile *file, uint32 backup_version);
-static void write_page_headers(BackupPageHeader2 *headers, pgFile *file, HeaderMap *hdr_map);
 static bool get_compressed_page_meta(FILE *in, const char *fullpath, BackupPageHeader* bph,
 									 pg_crc32 *crc, bool use_crc32c);
 
@@ -538,7 +536,7 @@ backup_data_file(ConnectionArgs* conn_arg, pgFile *file,
 				 XLogRecPtr prev_backup_start_lsn, BackupMode backup_mode,
 				 CompressAlg calg, int clevel, uint32 checksum_version,
 				 int ptrack_version_num, const char *ptrack_schema,
-				 HeaderMap *hdr_map, bool missing_ok)
+				 HeaderMap *hdr_map, bool is_merge)
 {
 	int         rc;
 	bool        use_pagemap;
@@ -629,7 +627,7 @@ backup_data_file(ConnectionArgs* conn_arg, pgFile *file,
 	/* check for errors */
 	if (rc == FILE_MISSING)
 	{
-		elog(LOG, "File \"%s\" is not found", from_fullpath);
+		elog(is_merge ? ERROR : LOG, "File not found: \"%s\"", from_fullpath);
 		file->write_size = FILE_NOT_FOUND;
 		goto cleanup;
 	}
@@ -685,7 +683,7 @@ backup_data_file(ConnectionArgs* conn_arg, pgFile *file,
 	FIN_FILE_CRC32(true, file->crc);
 
 	/* dump page headers */
-	write_page_headers(headers, file, hdr_map);
+	write_page_headers(headers, file, hdr_map, is_merge);
 
 	pg_free(errmsg);
 	pg_free(file->pagemap.bitmap);
@@ -818,8 +816,12 @@ restore_data_file(parray *parent_chain, pgFile *dest_file, FILE *out,
 		/* set stdio buffering for input data file */
 		setvbuf(in, in_buf, _IOFBF, STDIO_BUFSIZE);
 
+//		elog(INFO, "N_HEADERS: %i", tmp_file->n_headers);
+//		elog(INFO, "File: %s", tmp_file->rel_path);
+//		elog(INFO, "Backup: %s", base36enc(backup->start_time));
+
 		/* get headers for this file */
-		if (use_headers)
+		if (use_headers && tmp_file->n_headers > 0)
 			headers = get_data_file_headers(&(backup->hdr_map), tmp_file,
 											parse_program_version(backup->program_version));
 
@@ -2163,8 +2165,8 @@ get_data_file_headers(HeaderMap *hdr_map, pgFile *file, uint32 backup_version)
 	FIN_FILE_CRC32(true, hdr_crc);
 
 	if (hdr_crc != file->hdr_crc)
-		elog(ERROR, "Header file crc mismatch \"%s\", current: %u, expected: %u",
-			hdr_map->path, hdr_crc, file->hdr_crc);
+		elog(ERROR, "Header map for file \"%s\" crc mismatch \"%s\" offset: %lu, len: %lu, current: %u, expected: %u",
+			file->rel_path, hdr_map->path, file->hdr_off, read_len, hdr_crc, file->hdr_crc);
 
 	if (fclose(in))
 		elog(ERROR, "Cannot close header file \"%s\": %s", hdr_map->path, strerror(errno));
@@ -2173,29 +2175,35 @@ get_data_file_headers(HeaderMap *hdr_map, pgFile *file, uint32 backup_version)
 }
 
 void
-write_page_headers(BackupPageHeader2 *headers, pgFile *file, HeaderMap *hdr_map)
+write_page_headers(BackupPageHeader2 *headers, pgFile *file, HeaderMap *hdr_map, bool is_merge)
 {
-	size_t read_len = 0;
+	size_t  read_len = 0;
+	char   *map_path = NULL;
 
 	if (file->n_headers <= 0)
 		return;
 
+	/* when running merge we must save headers into the temp map */
+	map_path = (is_merge) ? hdr_map->path_tmp : hdr_map->path;
+
 	/* writing to header map must be serialized */
 	pthread_lock(&(hdr_map->mutex)); /* what if we crash while trying to obtain mutex? */
 
 	if (!hdr_map->fp)
 	{
-		hdr_map->fp = fopen(hdr_map->path, PG_BINARY_W);
+		elog(LOG, "Creating page header map \"%s\"", map_path);
+
+		hdr_map->fp = fopen(map_path, PG_BINARY_W);
 		if (hdr_map->fp == NULL)
 			elog(ERROR, "Cannot open header file \"%s\": %s",
-				 hdr_map->path, strerror(errno));
+				 map_path, strerror(errno));
 
 		/* disable buffering for header file */
 		setvbuf(hdr_map->fp, NULL, _IONBF, BUFSIZ);
 
 		/* update file permission */
-		if (chmod(hdr_map->path, FILE_PERMISSION) == -1)
-			elog(ERROR, "Cannot change mode of \"%s\": %s", hdr_map->path,
+		if (chmod(map_path, FILE_PERMISSION) == -1)
+			elog(ERROR, "Cannot change mode of \"%s\": %s", map_path,
 				 strerror(errno));
 
 		file->hdr_off = 0;
@@ -2211,7 +2219,10 @@ write_page_headers(BackupPageHeader2 *headers, pgFile *file, HeaderMap *hdr_map)
 	FIN_FILE_CRC32(true, file->hdr_crc);
 
 	if (fwrite(headers, 1, read_len, hdr_map->fp) != read_len)
-		elog(ERROR, "Cannot write to file \"%s\": %s", hdr_map->path, strerror(errno));
+		elog(ERROR, "Cannot write to file \"%s\": %s", map_path, strerror(errno));
+
+	elog(VERBOSE, "Writing header map for file \"%s\" offset: %lu, len: %lu, crc: %u",
+			file->rel_path, file->hdr_off, read_len, file->hdr_crc);
 
 	pthread_mutex_unlock(&(hdr_map->mutex));
 }

src/merge.c

@@ -550,11 +550,11 @@ merge_chain(parray *parent_chain, pgBackup *full_backup, pgBackup *dest_backup)
 	elog(INFO, "Validate parent chain for backup %s",
 					base36enc(dest_backup->start_time));
 
-	/* forbid merge retry for failed merges between 2.4.0 and any
+	/* Forbid merge retry for failed merges between 2.4.0 and any
 	 * older version. Several format changes makes it impossible
 	 * to determine the exact format any speific file is got.
 	 */
-	if (full_backup->status == BACKUP_STATUS_MERGING &&
+	if (is_retry &&
 		parse_program_version(dest_backup->program_version) >= 20400 &&
 		parse_program_version(full_backup->program_version) < 20400)
 	{
@@ -705,6 +705,23 @@ merge_chain(parray *parent_chain, pgBackup *full_backup, pgBackup *dest_backup)
 		elog(ERROR, "Backup files merging failed, time elapsed: %s",
 				pretty_time);
 
+	/* If temp header map descriptor is open, then close it and make rename */
+	if (full_backup->hdr_map.fp)
+	{
+		if (fclose(full_backup->hdr_map.fp))
+			elog(ERROR, "Cannot close file \"%s\"", full_backup->hdr_map.path);
+
+		/* sync new header map to dist */
+		if (fio_sync(full_backup->hdr_map.path_tmp, FIO_BACKUP_HOST) != 0)
+			elog(ERROR, "Cannot sync temp header map \"%s\": %s",
+				full_backup->hdr_map.path_tmp, strerror(errno));
+
+		/* Replace old header map with new one */
+		if (rename(full_backup->hdr_map.path_tmp, full_backup->hdr_map.path) == -1)
+			elog(ERROR, "Could not rename file \"%s\" to \"%s\": %s",
+				 full_backup->hdr_map.path_tmp, full_backup->hdr_map.path, strerror(errno));
+	}
+
 	/*
 	 * Update FULL backup metadata.
 	 * We cannot set backup status to OK just yet,
@@ -847,6 +864,12 @@ merge_chain(parray *parent_chain, pgBackup *full_backup, pgBackup *dest_backup)
 		full_backup->root_dir = pgut_strdup(destination_path);
 	}
 
+	/* Reinit some path variables */
+	join_path_components(full_backup->database_dir, full_backup->root_dir, DATABASE_DIR);
+	join_path_components(full_backup->hdr_map.path, full_backup->database_dir, HEADER_MAP);
+	join_path_components(full_backup->hdr_map.path_tmp, full_backup->database_dir, HEADER_MAP_TMP);
+	full_backup->hdr_map.fp = NULL;
+
 	/* If we crash here, it will produce full backup in MERGED
 	 * status, located in directory with wrong backup id.
 	 * It should not be a problem.
@@ -1033,6 +1056,7 @@ merge_files(void *arg)
 			if (file &&
 				file->n_blocks == dest_file->n_blocks)
 			{
+				BackupPageHeader2 *headers = NULL;
 
 				elog(VERBOSE, "The file didn`t changed since FULL backup, skip merge: \"%s\"",
 								file->rel_path);
@@ -1052,6 +1076,18 @@ merge_files(void *arg)
 				else
 					tmp_file->uncompressed_size = tmp_file->write_size;
 
+				/* Copy header metadata from old map into a new one */
+				tmp_file->n_headers = file->n_headers;
+				headers = get_data_file_headers(&(arguments->full_backup->hdr_map), file,
+						parse_program_version(arguments->full_backup->program_version));
+
+				/* sanity */
+				if (!headers && file->n_headers > 0)
+					elog(ERROR, "Failed to get headers for file \"%s\"", file->rel_path);
+
+				write_page_headers(headers, tmp_file, &(arguments->full_backup->hdr_map), true);
+				pg_free(headers);
+
 				//TODO: report in_place merge bytes.
 				goto done;
 			}
@@ -1205,16 +1241,14 @@ merge_data_file(parray *parent_chain, pgBackup *full_backup,
 	 * 16KB.
 	 * TODO: maybe we should just trust dest_file->n_blocks?
 	 * No, we can`t, because current binary can be used to merge
-	 * 2 backups of old versions, were n_blocks is missing.
+	 * 2 backups of old versions, where n_blocks is missing.
 	 */
 
 	backup_data_file(NULL, tmp_file, to_fullpath_tmp1, to_fullpath_tmp2,
 				 InvalidXLogRecPtr, BACKUP_MODE_FULL,
 				 dest_backup->compress_alg, dest_backup->compress_level,
 				 dest_backup->checksum_version, 0, NULL,
-
-				 /* TODO: add header map */
-				 NULL, false);
+				 &(full_backup->hdr_map), true);
 
 	/* drop restored temp file */
 	if (unlink(to_fullpath_tmp1) == -1)

src/pg_probackup.h

@@ -67,6 +67,7 @@ extern const char  *PROGRAM_EMAIL;
 #define EXTERNAL_DIR			"external_directories/externaldir"
 #define DATABASE_MAP			"database_map"
 #define HEADER_MAP  			"block_header_map"
+#define HEADER_MAP_TMP  		"block_header_map_tmp"
 
 /* Timeout defaults */
 #define ARCHIVE_TIMEOUT_DEFAULT		300
@@ -363,6 +364,7 @@ typedef struct PGNodeInfo
 typedef struct HeaderMap
 {
 	char  *path;
+	char  *path_tmp;	/* used only in merge */
 	FILE  *fp;
 	off_t  offset;
 	pthread_mutex_t mutex;
@@ -1014,6 +1016,9 @@ extern pid_t check_postmaster(const char *pgdata);
 
 extern bool validate_file_pages(pgFile *file, const char *fullpath, XLogRecPtr stop_lsn,
 							    uint32 checksum_version, uint32 backup_version, HeaderMap *hdr_map);
+
+extern BackupPageHeader2* get_data_file_headers(HeaderMap *hdr_map, pgFile *file, uint32 backup_version);
+extern void write_page_headers(BackupPageHeader2 *headers, pgFile *file, HeaderMap *hdr_map, bool is_merge);
 /* parsexlog.c */
 extern bool extractPageMap(const char *archivedir, uint32 wal_seg_size,
 						   XLogRecPtr startpoint, TimeLineID start_tli,