Commit ce6a71fa authored by Thomas Munro's avatar Thomas Munro

Use vectored I/O to fill new WAL segments.

Instead of making many block-sized write() calls to fill a new WAL file
with zeroes, make a smaller number of pwritev() calls (or various
emulations).  The actual number depends on the OS's IOV_MAX, which
PG_IOV_MAX currently caps at 32.  That means we'll write 256kB per call
on typical systems.  We may want to tune the number later with more
experience.
Reviewed-by: default avatarTom Lane <tgl@sss.pgh.pa.us>
Reviewed-by: default avatarAndres Freund <andres@anarazel.de>
Discussion: https://postgr.es/m/CA%2BhUKGJA%2Bu-220VONeoREBXJ9P3S94Y7J%2BkqCnTYmahvZJwM%3Dg%40mail.gmail.com
parent 13a021f3
...@@ -48,6 +48,7 @@ ...@@ -48,6 +48,7 @@
#include "pg_trace.h" #include "pg_trace.h"
#include "pgstat.h" #include "pgstat.h"
#include "port/atomics.h" #include "port/atomics.h"
#include "port/pg_iovec.h"
#include "postmaster/bgwriter.h" #include "postmaster/bgwriter.h"
#include "postmaster/startup.h" #include "postmaster/startup.h"
#include "postmaster/walwriter.h" #include "postmaster/walwriter.h"
...@@ -3270,7 +3271,6 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock) ...@@ -3270,7 +3271,6 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock)
XLogSegNo installed_segno; XLogSegNo installed_segno;
XLogSegNo max_segno; XLogSegNo max_segno;
int fd; int fd;
int nbytes;
int save_errno; int save_errno;
XLogFilePath(path, ThisTimeLineID, logsegno, wal_segment_size); XLogFilePath(path, ThisTimeLineID, logsegno, wal_segment_size);
...@@ -3317,6 +3317,9 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock) ...@@ -3317,6 +3317,9 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock)
save_errno = 0; save_errno = 0;
if (wal_init_zero) if (wal_init_zero)
{ {
struct iovec iov[PG_IOV_MAX];
int blocks;
/* /*
* Zero-fill the file. With this setting, we do this the hard way to * Zero-fill the file. With this setting, we do this the hard way to
* ensure that all the file space has really been allocated. On * ensure that all the file space has really been allocated. On
...@@ -3326,15 +3329,28 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock) ...@@ -3326,15 +3329,28 @@ XLogFileInit(XLogSegNo logsegno, bool *use_existent, bool use_lock)
* indirect blocks are down on disk. Therefore, fdatasync(2) or * indirect blocks are down on disk. Therefore, fdatasync(2) or
* O_DSYNC will be sufficient to sync future writes to the log file. * O_DSYNC will be sufficient to sync future writes to the log file.
*/ */
for (nbytes = 0; nbytes < wal_segment_size; nbytes += XLOG_BLCKSZ)
/* Prepare to write out a lot of copies of our zero buffer at once. */
for (int i = 0; i < lengthof(iov); ++i)
{ {
errno = 0; iov[i].iov_base = zbuffer.data;
if (write(fd, zbuffer.data, XLOG_BLCKSZ) != XLOG_BLCKSZ) iov[i].iov_len = XLOG_BLCKSZ;
}
/* Loop, writing as many blocks as we can for each system call. */
blocks = wal_segment_size / XLOG_BLCKSZ;
for (int i = 0; i < blocks;)
{ {
/* if write didn't set errno, assume no disk space */ int iovcnt = Min(blocks - i, lengthof(iov));
save_errno = errno ? errno : ENOSPC; off_t offset = i * XLOG_BLCKSZ;
if (pg_pwritev_with_retry(fd, iov, iovcnt, offset) < 0)
{
save_errno = errno;
break; break;
} }
i += iovcnt;
} }
} }
else else
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment