pg_basebackup 不备份哪些文件
背景
pg_basebackup是PostgreSQL提供的基础备份工具，用户可以使用这个工具来实现对数据库的基础备份，当然你也可以使用pg_start_backup()+copy+pg_stop_backup()的方式进行备份。
那么pg_basebackup会帮你过滤哪些文件，或者不备份哪些文件呢？
1. pgssql_tmp开头的临时目录中的文件
2. postgresql.auto.conf.tmp
3. BACKUP_LABEL_FILE
4. postmaster.pid
5. pg_stat_tmp目录中的文件
6. pg_replslot目录中的文件
7. 如果没有加-x参数则同样跳过pg_xlog中的文件
源码如下：
src/include/storage/fd.h:#define PG_TEMP_FILE_PREFIX "pgsql_tmp"
src/include/utils/guc.h:#define PG_AUTOCONF_FILENAME "postgresql.auto.conf"
contrib/pg_stat_statements/pg_stat_statements.c:#define PGSS_TEXT_FILE PG_STAT_TMP_DIR "/pgss_query_texts.stat"
src/include/pgstat.h:#define PG_STAT_TMP_DIR "pg_stat_tmp"
src/backend/replication/basebackup.c
/*
* Include all files from the given directory in the output tar stream. If
* 'sizeonly' is true, we just calculate a total length and return it, without
* actually sending anything.
*
* Omit any directory in the tablespaces list, to avoid backing up
* tablespaces twice when they were created inside PGDATA.
*/
static int64
sendDir(char *path, int basepathlen, bool sizeonly, List *tablespaces)
{
......
dir = AllocateDir(path);
while ((de = ReadDir(dir, path)) != NULL)
{
/* Skip special stuff */
if (strcmp(de->d_name, ".") == 0 || strcmp(de->d_name, "..") == 0)
continue;
跳过临时文件，即pgsql_tmp开头的文件
/* Skip temporary files */
if (strncmp(de->d_name,
PG_TEMP_FILE_PREFIX,
strlen(PG_TEMP_FILE_PREFIX)) == 0)
continue;
跳过postgresql.auto.conf的临时文件postgresql.auto.conf.tmp
/* skip auto conf temporary file */
if (strncmp(de->d_name,
PG_AUTOCONF_FILENAME ".tmp",
sizeof(PG_AUTOCONF_FILENAME) + 4) == 0)
continue;
跳过使用pg_start_backup()+copy+pg_stop_backup()的方式进行备份的标签文件，BACKUP_LABEL_FILE
/*
* If there's a backup_label file, it belongs to a backup started by
* the user with pg_start_backup(). It is *not* correct for this
* backup, our backup_label is injected into the tar separately.
*/
if (strcmp(de->d_name, BACKUP_LABEL_FILE) == 0)
continue;
......
跳过pid文件
/* Skip postmaster.pid and postmaster.opts in the data directory */
if (strcmp(pathbuf, "./postmaster.pid") == 0 ||
strcmp(pathbuf, "./postmaster.opts") == 0)
continue;
跳过控制文件，不是不备份，是放在最后备份，因为使用pg_basebackup不产生标签文件来记录备份开始的XLOG OFFSET，所以需要将pg_control文件放在最后备份。
/* Skip pg_control here to back up it last */
if (strcmp(pathbuf, "./global/pg_control") == 0)
continue;
......
跳过统计信息的临时文件（包括pg_stat_statements插件的，因为这个插件的临时统计信息也放在pg_stat_tmp目录下。），
/*
* Skip temporary statistics files. PG_STAT_TMP_DIR must be skipped
* even when stats_temp_directory is set because PGSS_TEXT_FILE is
* always created there.
*/
if ((statrelpath != NULL && strcmp(pathbuf, statrelpath) == 0) ||
strncmp(de->d_name, PG_STAT_TMP_DIR, strlen(PG_STAT_TMP_DIR)) == 0)
{
if (!sizeonly)
_tarWriteHeader(pathbuf + basepathlen + 1, NULL, &statbuf);
size += 512;
continue;
}
跳过pg_replslot目录中的文件
/*
* Skip pg_replslot, not useful to copy. But include it as an empty
* directory anyway, so we get permissions right.
*/
if (strcmp(de->d_name, "pg_replslot") == 0)
{
if (!sizeonly)
_tarWriteHeader(pathbuf + basepathlen + 1, NULL, &statbuf);
size += 512; /* Size of the header just added */
continue;
}
......
跳过pg_xlog，
当打开了-x参数时，在另一处代码备份xlog
-x, --xlog include required WAL files in backup (fetch mode)
/*
* We can skip pg_xlog, the WAL segments need to be fetched from the
* WAL archive anyway. But include it as an empty directory anyway, so
* we get permissions right.
*/
if (strcmp(pathbuf, "./pg_xlog") == 0)
{
if (!sizeonly)
{
/* If pg_xlog is a symlink, write it as a directory anyway */
#ifndef WIN32
if (S_ISLNK(statbuf.st_mode))
#else
if (pgwin32_is_junction(pathbuf))
#endif
statbuf.st_mode = S_IFDIR | S_IRWXU;
_tarWriteHeader(pathbuf + basepathlen + 1, NULL, &statbuf);
}
size += 512; /* Size of the header just added */
/*
* Also send archive_status directory (by hackishly reusing
* statbuf from above ...).
*/
if (!sizeonly)
_tarWriteHeader("./pg_xlog/archive_status", NULL, &statbuf);
size += 512; /* Size of the header just added */
continue; /* don't recurse into pg_xlog */
}
当打开了-x参数时，在此处代码备份xlog
src/backend/replication/basebackup.c
/*
* Actually do a base backup for the specified tablespaces.
*
* This is split out mainly to avoid complaints about "variable might be
* clobbered by longjmp" from stupider versions of gcc.
*/
static void
perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
{
......
if (opt->includewal)
{
/*
......
最后，我们注意到pg_basebackup备份了hash index，以及unlogged table的内容，实际上他们的备份都是无用功，因为他们不产生XLOG，在还原时，unlogged table会自动清掉，而hash index则可能存在不一致数据。
这是不合理的地方。
参考
1. src/backend/replication/basebackup.c