Hi,
On 2014-12-05 16:18:02 +0900, Fujii Masao wrote:
> On Fri, Dec 5, 2014 at 9:28 AM, Andres Freund <and...@2ndquadrant.com> wrote:
> > So I think we just need to make pg_basebackup create to .ready
> > files.
>
> s/.ready/.done? If yes, +1.
That unfortunately requires changes to both backend and pg_basebackup to
support fetch and stream modes respectively.
I've attached a preliminary patch for this. I'd appreciate feedback. I
plan to commit it in a couple of days, after some more
testing/rereading.
> > Given that the walreceiver and restore_command already
> > unconditionally do XLogArchiveForceDone() I think we'd follow the
> > established precedent. Arguably it could make sense to archive files
> > again on the standby after a promotion as they aren't guaranteed to have
> > been on the then primary. But we don't have any infrastructure anyway
> > for that and walsender doesn't do so, so it doesn't seem to make any
> > sense to do that for pg_basebackup.
> >
> > Independent from this bug, there's also some debatable behaviour about
> > what happens if a node with a high wal_keep_segments turns on
> > archive_mode. Suddenly all those old files are archived... I think it
> > might be a good idea to simply always create .done files when
> > archive_mode is disabled while a wal segment is finished.
>
> +1
I tend to think that's a master only change. Agreed?
Greetings,
Andres Freund
>From 3db116bc5b9465f555957bb11ac6cb8b20c18405 Mon Sep 17 00:00:00 2001
From: Andres Freund <and...@anarazel.de>
Date: Wed, 31 Dec 2014 14:50:57 +0100
Subject: [PATCH 1/2] Add pg_string_endswith as the start of a string helper
library in src/common.
Backpatch to 9.3 where src/common was introduce, because a bugfix that
needs to be backpatched, requires the function. Earlier branches will
have to duplicate the code.
---
src/backend/replication/slot.c | 21 ++-------------------
src/common/Makefile | 2 +-
src/common/string.c | 43 ++++++++++++++++++++++++++++++++++++++++++
src/include/common/string.h | 15 +++++++++++++++
src/tools/msvc/Mkvcbuild.pm | 2 +-
5 files changed, 62 insertions(+), 21 deletions(-)
create mode 100644 src/common/string.c
create mode 100644 src/include/common/string.h
diff --git a/src/backend/replication/slot.c b/src/backend/replication/slot.c
index 937b669..698ca6b 100644
--- a/src/backend/replication/slot.c
+++ b/src/backend/replication/slot.c
@@ -40,6 +40,7 @@
#include <sys/stat.h>
#include "access/transam.h"
+#include "common/string.h"
#include "miscadmin.h"
#include "replication/slot.h"
#include "storage/fd.h"
@@ -780,24 +781,6 @@ CheckSlotRequirements(void)
}
/*
- * Returns whether the string `str' has the postfix `end'.
- */
-static bool
-string_endswith(const char *str, const char *end)
-{
- size_t slen = strlen(str);
- size_t elen = strlen(end);
-
- /* can't be a postfix if longer */
- if (elen > slen)
- return false;
-
- /* compare the end of the strings */
- str += slen - elen;
- return strcmp(str, end) == 0;
-}
-
-/*
* Flush all replication slots to disk.
*
* This needn't actually be part of a checkpoint, but it's a convenient
@@ -864,7 +847,7 @@ StartupReplicationSlots(void)
continue;
/* we crashed while a slot was being setup or deleted, clean up */
- if (string_endswith(replication_de->d_name, ".tmp"))
+ if (pg_string_endswith(replication_de->d_name, ".tmp"))
{
if (!rmtree(path, true))
{
diff --git a/src/common/Makefile b/src/common/Makefile
index 7edbaaa..e5c345d 100644
--- a/src/common/Makefile
+++ b/src/common/Makefile
@@ -23,7 +23,7 @@ include $(top_builddir)/src/Makefile.global
override CPPFLAGS := -DFRONTEND $(CPPFLAGS)
LIBS += $(PTHREAD_LIBS)
-OBJS_COMMON = exec.o pgfnames.o psprintf.o relpath.o rmtree.o username.o wait_error.o
+OBJS_COMMON = exec.o pgfnames.o psprintf.o relpath.o rmtree.o string.o username.o wait_error.o
OBJS_FRONTEND = $(OBJS_COMMON) fe_memutils.o
diff --git a/src/common/string.c b/src/common/string.c
new file mode 100644
index 0000000..07a2aaf
--- /dev/null
+++ b/src/common/string.c
@@ -0,0 +1,43 @@
+/*-------------------------------------------------------------------------
+ *
+ * string.c
+ * string handling helpers
+ *
+ *
+ * Portions Copyright (c) 1996-2014, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ *
+ * IDENTIFICATION
+ * src/common/string.c
+ *
+ *-------------------------------------------------------------------------
+ */
+
+
+#ifndef FRONTEND
+#include "postgres.h"
+#else
+#include "postgres_fe.h"
+#endif
+
+#include "common/string.h"
+
+
+/*
+ * Returns whether the string `str' has the postfix `end'.
+ */
+bool
+pg_string_endswith(const char *str, const char *end)
+{
+ size_t slen = strlen(str);
+ size_t elen = strlen(end);
+
+ /* can't be a postfix if longer */
+ if (elen > slen)
+ return false;
+
+ /* compare the end of the strings */
+ str += slen - elen;
+ return strcmp(str, end) == 0;
+}
diff --git a/src/include/common/string.h b/src/include/common/string.h
new file mode 100644
index 0000000..3e1a650
--- /dev/null
+++ b/src/include/common/string.h
@@ -0,0 +1,15 @@
+/*
+ * string.h
+ * string handling helpers
+ *
+ * Portions Copyright (c) 1996-2014, PostgreSQL Global Development Group
+ * Portions Copyright (c) 1994, Regents of the University of California
+ *
+ * src/include/common/string.h
+ */
+#ifndef COMMON_STRING_H
+#define COMMON_STRING_H
+
+extern bool pg_string_endswith(const char *str, const char *end);
+
+#endif /* COMMON_STRING_H */
diff --git a/src/tools/msvc/Mkvcbuild.pm b/src/tools/msvc/Mkvcbuild.pm
index 4506739..4336f2e 100644
--- a/src/tools/msvc/Mkvcbuild.pm
+++ b/src/tools/msvc/Mkvcbuild.pm
@@ -76,7 +76,7 @@ sub mkvcbuild
push(@pgportfiles, 'rint.c') if ($vsVersion < '12.00');
our @pgcommonallfiles = qw(
- exec.c pgfnames.c psprintf.c relpath.c rmtree.c username.c wait_error.c);
+ exec.c pgfnames.c psprintf.c relpath.c rmtree.c string.c username.c wait_error.c);
our @pgcommonfrontendfiles = (@pgcommonallfiles, qw(fe_memutils.c));
--
2.2.0.rc0.18.ga1ad247
>From 6cad7b85c4fb36c2c5675bb83070693a0a7b9232 Mon Sep 17 00:00:00 2001
From: Andres Freund <and...@anarazel.de>
Date: Wed, 31 Dec 2014 15:40:13 +0100
Subject: [PATCH 2/2] Prevent WAL files created by pg_basebackup -x/X from
being archived again.
WAL (and timeline history) files created by pg_basebackup did not
maintain the new base backup's archive status. That's currently not a
problem if the new node is used as a primary - but if that node is
promoted all still existing files can get archived again. With a high
wal_keep_segment settings that can be a significant time later - which
is quite confusing.
Change both the backend (for the -x fetch case) and pg_basebackup
itself to always mark files as .done. That's in line with
walreceiver.c doing so.
Backpatch to 9.1 where pg_basebackup was introduced.
---
src/backend/replication/basebackup.c | 24 ++++++++++
src/bin/pg_basebackup/pg_basebackup.c | 32 +++++++++----
src/bin/pg_basebackup/pg_receivexlog.c | 2 +-
src/bin/pg_basebackup/receivelog.c | 87 ++++++++++++++++++++++++++--------
src/bin/pg_basebackup/receivelog.h | 3 +-
5 files changed, 117 insertions(+), 31 deletions(-)
diff --git a/src/backend/replication/basebackup.c b/src/backend/replication/basebackup.c
index fbcecbb..24c3d8d 100644
--- a/src/backend/replication/basebackup.c
+++ b/src/backend/replication/basebackup.c
@@ -471,6 +471,7 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
errmsg("unexpected WAL file size \"%s\"", walFiles[i])));
}
+ /* send the WAL file itself */
_tarWriteHeader(pathbuf, NULL, &statbuf);
while ((cnt = fread(buf, 1, Min(sizeof(buf), XLogSegSize - len), fp)) > 0)
@@ -497,7 +498,17 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
}
/* XLogSegSize is a multiple of 512, so no need for padding */
+
FreeFile(fp);
+
+ /*
+ * Mark file as archived, otherwise files can get archived again
+ * after promotion of a new node. This is in line with
+ * walreceiver.c always doing a XLogArchiveForceDone() after a
+ * complete segment.
+ */
+ StatusFilePath(pathbuf, walFiles[i], ".done");
+ sendFileWithContent(pathbuf, "");
}
/*
@@ -521,6 +532,10 @@ perform_base_backup(basebackup_options *opt, DIR *tblspcdir)
errmsg("could not stat file \"%s\": %m", pathbuf)));
sendFile(pathbuf, pathbuf, &statbuf, false);
+
+ /* unconditionally mark file as archived */
+ StatusFilePath(pathbuf, fname, ".done");
+ sendFileWithContent(pathbuf, "");
}
/* Send CopyDone message for the last tar file */
@@ -1021,6 +1036,15 @@ sendDir(char *path, int basepathlen, bool sizeonly, List *tablespaces)
_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 */
}
diff --git a/src/bin/pg_basebackup/pg_basebackup.c b/src/bin/pg_basebackup/pg_basebackup.c
index 0470401..fc30a3c 100644
--- a/src/bin/pg_basebackup/pg_basebackup.c
+++ b/src/bin/pg_basebackup/pg_basebackup.c
@@ -25,6 +25,7 @@
#include <zlib.h>
#endif
+#include "common/string.h"
#include "getopt_long.h"
#include "libpq-fe.h"
#include "pqexpbuffer.h"
@@ -370,7 +371,7 @@ LogStreamerMain(logstreamer_param *param)
if (!ReceiveXlogStream(param->bgconn, param->startptr, param->timeline,
param->sysidentifier, param->xlogdir,
reached_end_position, standby_message_timeout,
- NULL, false))
+ NULL, false, true))
/*
* Any errors will already have been reported in the function process,
@@ -394,6 +395,7 @@ StartLogStreamer(char *startpos, uint32 timeline, char *sysidentifier)
logstreamer_param *param;
uint32 hi,
lo;
+ char *statusdir;
param = pg_malloc0(sizeof(logstreamer_param));
param->timeline = timeline;
@@ -428,13 +430,24 @@ StartLogStreamer(char *startpos, uint32 timeline, char *sysidentifier)
/* Error message already written in GetConnection() */
exit(1);
+ snprintf(param->xlogdir, sizeof(param->xlogdir), "%s/pg_xlog", basedir);
+
/*
- * Always in plain format, so we can write to basedir/pg_xlog. But the
- * directory entry in the tar file may arrive later, so make sure it's
- * created before we start.
+ * Create pg_xlog/archive_status (and thus pg_xlog) so we can can write to
+ * basedir/pg_xlog as the directory entry in the tar file may arrive
+ * later.
*/
- snprintf(param->xlogdir, sizeof(param->xlogdir), "%s/pg_xlog", basedir);
- verify_dir_is_empty_or_create(param->xlogdir);
+ statusdir = psprintf("%s/pg_xlog/archive_status", basedir);
+
+ if (pg_mkdir_p(statusdir, S_IRWXU) != 0 && errno != EEXIST)
+ {
+ fprintf(stderr,
+ _("%s: could not create directory \"%s\": %s\n"),
+ progname, param->xlogdir, strerror(errno));
+ disconnect_and_exit(1);
+ }
+
+ free(statusdir);
/*
* Start a child process and tell it to start streaming. On Unix, this is
@@ -1237,10 +1250,11 @@ ReceiveAndUnpackTarFile(PGconn *conn, PGresult *res, int rownum)
* log directory location was specified, pg_xlog has
* already been created as a symbolic link before
* starting the actual backup. So just ignore failure
- * on them.
+ * on related directories.
*/
- if ((!streamwal && (strcmp(xlog_dir, "") == 0))
- || strcmp(filename + strlen(filename) - 8, "/pg_xlog") != 0)
+ if (errno == EEXIST &&
+ !pg_string_endswith(filename, "/pg_xlog") &&
+ !pg_string_endswith(filename, "/archive_status"))
{
fprintf(stderr,
_("%s: could not create directory \"%s\": %s\n"),
diff --git a/src/bin/pg_basebackup/pg_receivexlog.c b/src/bin/pg_basebackup/pg_receivexlog.c
index 4658f08..b10da73 100644
--- a/src/bin/pg_basebackup/pg_receivexlog.c
+++ b/src/bin/pg_basebackup/pg_receivexlog.c
@@ -342,7 +342,7 @@ StreamLog(void)
ReceiveXlogStream(conn, startpos, starttli, NULL, basedir,
stop_streaming, standby_message_timeout, ".partial",
- synchronous);
+ synchronous, false);
PQfinish(conn);
conn = NULL;
diff --git a/src/bin/pg_basebackup/receivelog.c b/src/bin/pg_basebackup/receivelog.c
index f0f8760..5ce1c7a 100644
--- a/src/bin/pg_basebackup/receivelog.c
+++ b/src/bin/pg_basebackup/receivelog.c
@@ -37,7 +37,7 @@ static PGresult *HandleCopyStream(PGconn *conn, XLogRecPtr startpos,
uint32 timeline, char *basedir,
stream_stop_callback stream_stop, int standby_message_timeout,
char *partial_suffix, XLogRecPtr *stoppos,
- bool synchronous);
+ bool synchronous, bool mark_done);
static int CopyStreamPoll(PGconn *conn, long timeout_ms);
static int CopyStreamReceive(PGconn *conn, long timeout, char **buffer);
static bool ProcessKeepaliveMsg(PGconn *conn, char *copybuf, int len,
@@ -45,20 +45,40 @@ static bool ProcessKeepaliveMsg(PGconn *conn, char *copybuf, int len,
static bool ProcessXLogDataMsg(PGconn *conn, char *copybuf, int len,
XLogRecPtr *blockpos, uint32 timeline,
char *basedir, stream_stop_callback stream_stop,
- char *partial_suffix);
+ char *partial_suffix, bool mark_done);
static PGresult *HandleEndOfCopyStream(PGconn *conn, char *copybuf,
XLogRecPtr blockpos, char *basedir, char *partial_suffix,
- XLogRecPtr *stoppos);
+ XLogRecPtr *stoppos, bool mark_done);
static bool CheckCopyStreamStop(PGconn *conn, XLogRecPtr blockpos,
uint32 timeline, char *basedir,
stream_stop_callback stream_stop,
- char *partial_suffix, XLogRecPtr *stoppos);
+ char *partial_suffix, XLogRecPtr *stoppos,
+ bool mark_done);
static long CalculateCopyStreamSleeptime(int64 now, int standby_message_timeout,
int64 last_status);
static bool ReadEndOfStreamingResult(PGresult *res, XLogRecPtr *startpos,
uint32 *timeline);
+static bool
+mark_file_as_archived(const char *basedir, const char *fname)
+{
+ int fd;
+ static char tmppath[MAXPGPATH];
+
+ snprintf(tmppath, sizeof(tmppath), "%s/archive_status/%s.done",
+ basedir, fname);
+
+ fd = open(tmppath, O_WRONLY | O_CREAT | PG_BINARY, S_IRUSR | S_IWUSR);
+ if (fd < 0)
+ {
+ fprintf(stderr, _("%s: could not create archive status file \"%s\": %s\n"),
+ progname, tmppath, strerror(errno));
+ return false;
+ }
+ return false;
+}
+
/*
* Open a new WAL file in the specified directory.
*
@@ -152,7 +172,7 @@ open_walfile(XLogRecPtr startpoint, uint32 timeline, char *basedir,
* and returns false, otherwise returns true.
*/
static bool
-close_walfile(char *basedir, char *partial_suffix, XLogRecPtr pos)
+close_walfile(char *basedir, char *partial_suffix, XLogRecPtr pos, bool mark_done)
{
off_t currpos;
@@ -206,6 +226,19 @@ close_walfile(char *basedir, char *partial_suffix, XLogRecPtr pos)
_("%s: not renaming \"%s%s\", segment is not complete\n"),
progname, current_walfile_name, partial_suffix);
+ /*
+ * Mark file as archived if requested by the caller - pg_basebackup needs
+ * to do so as files can otherwise get archived again after promotion of a
+ * new node. This is in line with walreceiver.c always doing a
+ * XLogArchiveForceDone() after a complete segment.
+ */
+ if (currpos == XLOG_SEG_SIZE && mark_done)
+ {
+ /* writes error message if failed */
+ if (!mark_file_as_archived(basedir, current_walfile_name))
+ return false;
+ }
+
lastFlushPosition = pos;
return true;
}
@@ -248,7 +281,8 @@ existsTimeLineHistoryFile(char *basedir, TimeLineID tli)
}
static bool
-writeTimeLineHistoryFile(char *basedir, TimeLineID tli, char *filename, char *content)
+writeTimeLineHistoryFile(char *basedir, TimeLineID tli, char *filename,
+ char *content, bool mark_done)
{
int size = strlen(content);
char path[MAXPGPATH];
@@ -327,6 +361,14 @@ writeTimeLineHistoryFile(char *basedir, TimeLineID tli, char *filename, char *co
return false;
}
+ /* Maintain archive_status, check close_walfile() for details. */
+ if (mark_done)
+ {
+ /* writes error message if failed */
+ if (!mark_file_as_archived(basedir, histfname))
+ return false;
+ }
+
return true;
}
@@ -447,7 +489,7 @@ ReceiveXlogStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
char *sysidentifier, char *basedir,
stream_stop_callback stream_stop,
int standby_message_timeout, char *partial_suffix,
- bool synchronous)
+ bool synchronous, bool mark_done)
{
char query[128];
char slotcmd[128];
@@ -562,7 +604,8 @@ ReceiveXlogStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
/* Write the history file to disk */
writeTimeLineHistoryFile(basedir, timeline,
PQgetvalue(res, 0, 0),
- PQgetvalue(res, 0, 1));
+ PQgetvalue(res, 0, 1),
+ mark_done);
PQclear(res);
}
@@ -592,7 +635,7 @@ ReceiveXlogStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
/* Stream the WAL */
res = HandleCopyStream(conn, startpos, timeline, basedir, stream_stop,
standby_message_timeout, partial_suffix,
- &stoppos, synchronous);
+ &stoppos, synchronous, mark_done);
if (res == NULL)
goto error;
@@ -757,7 +800,7 @@ static PGresult *
HandleCopyStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
char *basedir, stream_stop_callback stream_stop,
int standby_message_timeout, char *partial_suffix,
- XLogRecPtr *stoppos, bool synchronous)
+ XLogRecPtr *stoppos, bool synchronous, bool mark_done)
{
char *copybuf = NULL;
int64 last_status = -1;
@@ -775,7 +818,8 @@ HandleCopyStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
* Check if we should continue streaming, or abort at this point.
*/
if (!CheckCopyStreamStop(conn, blockpos, timeline, basedir,
- stream_stop, partial_suffix, stoppos))
+ stream_stop, partial_suffix, stoppos,
+ mark_done))
goto error;
now = feGetCurrentTimestamp();
@@ -830,7 +874,8 @@ HandleCopyStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
if (r == -2)
{
PGresult *res = HandleEndOfCopyStream(conn, copybuf, blockpos,
- basedir, partial_suffix, stoppos);
+ basedir, partial_suffix,
+ stoppos, mark_done);
if (res == NULL)
goto error;
else
@@ -847,14 +892,16 @@ HandleCopyStream(PGconn *conn, XLogRecPtr startpos, uint32 timeline,
else if (copybuf[0] == 'w')
{
if (!ProcessXLogDataMsg(conn, copybuf, r, &blockpos,
- timeline, basedir, stream_stop, partial_suffix))
+ timeline, basedir, stream_stop,
+ partial_suffix, true))
goto error;
/*
* Check if we should continue streaming, or abort at this point.
*/
if (!CheckCopyStreamStop(conn, blockpos, timeline, basedir,
- stream_stop, partial_suffix, stoppos))
+ stream_stop, partial_suffix, stoppos,
+ mark_done))
goto error;
}
else
@@ -1055,7 +1102,7 @@ static bool
ProcessXLogDataMsg(PGconn *conn, char *copybuf, int len,
XLogRecPtr *blockpos, uint32 timeline,
char *basedir, stream_stop_callback stream_stop,
- char *partial_suffix)
+ char *partial_suffix, bool mark_done)
{
int xlogoff;
int bytes_left;
@@ -1163,7 +1210,7 @@ ProcessXLogDataMsg(PGconn *conn, char *copybuf, int len,
/* Did we reach the end of a WAL segment? */
if (*blockpos % XLOG_SEG_SIZE == 0)
{
- if (!close_walfile(basedir, partial_suffix, *blockpos))
+ if (!close_walfile(basedir, partial_suffix, *blockpos, mark_done))
/* Error message written in close_walfile() */
return false;
@@ -1193,7 +1240,7 @@ ProcessXLogDataMsg(PGconn *conn, char *copybuf, int len,
static PGresult *
HandleEndOfCopyStream(PGconn *conn, char *copybuf,
XLogRecPtr blockpos, char *basedir, char *partial_suffix,
- XLogRecPtr *stoppos)
+ XLogRecPtr *stoppos, bool mark_done)
{
PGresult *res = PQgetResult(conn);
@@ -1204,7 +1251,7 @@ HandleEndOfCopyStream(PGconn *conn, char *copybuf,
*/
if (still_sending)
{
- if (!close_walfile(basedir, partial_suffix, blockpos))
+ if (!close_walfile(basedir, partial_suffix, blockpos, mark_done))
{
/* Error message written in close_walfile() */
PQclear(res);
@@ -1236,11 +1283,11 @@ HandleEndOfCopyStream(PGconn *conn, char *copybuf,
static bool
CheckCopyStreamStop(PGconn *conn, XLogRecPtr blockpos, uint32 timeline,
char *basedir, stream_stop_callback stream_stop,
- char *partial_suffix, XLogRecPtr *stoppos)
+ char *partial_suffix, XLogRecPtr *stoppos, bool mark_done)
{
if (still_sending && stream_stop(blockpos, timeline, false))
{
- if (!close_walfile(basedir, partial_suffix, blockpos))
+ if (!close_walfile(basedir, partial_suffix, blockpos, mark_done))
{
/* Potential error message is written by close_walfile */
return false;
diff --git a/src/bin/pg_basebackup/receivelog.h b/src/bin/pg_basebackup/receivelog.h
index 9dd7005..1f64a74 100644
--- a/src/bin/pg_basebackup/receivelog.h
+++ b/src/bin/pg_basebackup/receivelog.h
@@ -31,6 +31,7 @@ extern bool ReceiveXlogStream(PGconn *conn,
stream_stop_callback stream_stop,
int standby_message_timeout,
char *partial_suffix,
- bool synchronous);
+ bool synchronous,
+ bool mark_done);
#endif /* RECEIVELOG_H */
--
2.2.0.rc0.18.ga1ad247
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
