On Monday 28 December 2009 23:54:51 Andres Freund wrote:
> On Saturday 12 December 2009 21:38:41 Andres Freund wrote:
> > On Saturday 12 December 2009 21:36:27 Michael Clemmons wrote:
> > > If ppl think its worth it I'll create a ticket
> >
> > Thanks, no need. I will post a patch tomorrow or so.
>
> Well. It was a long day...
>
> Anyway.
> In this patch I delay the fsync done in copy_file and simply do a second
> pass over the directory in copy_dir and fsync everything in that pass.
> Including the directory - which was not done before and actually might be
> necessary in some cases.
> I added a posix_fadvise(..., FADV_DONTNEED) to make it more likely that the
> copied file reaches storage before the fsync. Without the speed benefits
> were quite a bit smaller and essentially random (which seems sensible).
>
> This speeds up CREATE DATABASE from ~9 seconds to something around 0.8s on
> my laptop. Still slower than with fsync off (~0.25) but quite a worthy
> improvement.
>
> The benefits are obviously bigger if the template database includes
> anything added.
Obviously the patch would be helpfull.
Andres
From bd80748883d1328a71607a447677b0bfb1f54ab0 Mon Sep 17 00:00:00 2001
From: Andres Freund <andres@xxxxxxxxxxx>
Date: Mon, 28 Dec 2009 23:43:57 +0100
Subject: [PATCH] Delay fsyncing files during copying in CREATE DATABASE - this
dramatically speeds up CREATE DATABASE on non battery backed
rotational storage.
Additionally fsync() the directory to ensure all metadata reaches
storage.
---
src/port/copydir.c | 58 +++++++++++++++++++++++++++++++++++++++++++++------
1 files changed, 51 insertions(+), 7 deletions(-)
diff --git a/src/port/copydir.c b/src/port/copydir.c
index a70477e..cde3dc7 100644
*** a/src/port/copydir.c
--- b/src/port/copydir.c
***************
*** 37,42 ****
--- 37,43 ----
static void copy_file(char *fromfile, char *tofile);
+ static void fsync_fname(char *fname);
/*
*************** copydir(char *fromdir, char *todir, bool
*** 64,69 ****
--- 65,73 ----
(errcode_for_file_access(),
errmsg("could not open directory \"%s\": %m", fromdir)));
+ /*
+ * Copy all the files
+ */
while ((xlde = ReadDir(xldir, fromdir)) != NULL)
{
struct stat fst;
*************** copydir(char *fromdir, char *todir, bool
*** 89,96 ****
else if (S_ISREG(fst.st_mode))
copy_file(fromfile, tofile);
}
-
FreeDir(xldir);
}
/*
--- 93,120 ----
else if (S_ISREG(fst.st_mode))
copy_file(fromfile, tofile);
}
FreeDir(xldir);
+
+ /*
+ * Be paranoid here and fsync all files to ensure we catch problems.
+ */
+ xldir = AllocateDir(fromdir);
+ if (xldir == NULL)
+ ereport(ERROR,
+ (errcode_for_file_access(),
+ errmsg("could not open directory \"%s\": %m", fromdir)));
+
+ while ((xlde = ReadDir(xldir, fromdir)) != NULL)
+ {
+ struct stat fst;
+
+ if (strcmp(xlde->d_name, ".") == 0 ||
+ strcmp(xlde->d_name, "..") == 0)
+ continue;
+
+ snprintf(tofile, MAXPGPATH, "%s/%s", todir, xlde->d_name);
+ fsync_fname(tofile);
+ }
}
/*
*************** copy_file(char *fromfile, char *tofile)
*** 150,162 ****
}
/*
! * Be paranoid here to ensure we catch problems.
*/
! if (pg_fsync(dstfd) != 0)
! ereport(ERROR,
! (errcode_for_file_access(),
! errmsg("could not fsync file \"%s\": %m", tofile)));
!
if (close(dstfd))
ereport(ERROR,
(errcode_for_file_access(),
--- 174,185 ----
}
/*
! * We tell the kernel here to write the data back in order to make
! * the later fsync cheaper.
*/
! #if defined(USE_POSIX_FADVISE) && defined(POSIX_FADV_DONTNEED)
! posix_fadvise(dstfd, 0, 0, POSIX_FADV_DONTNEED);
! #endif
if (close(dstfd))
ereport(ERROR,
(errcode_for_file_access(),
*************** copy_file(char *fromfile, char *tofile)
*** 166,168 ****
--- 189,212 ----
pfree(buffer);
}
+
+ /*
+ * fsync a file
+ */
+ static void
+ fsync_fname(char *fname)
+ {
+ int fd = BasicOpenFile(fname, O_RDWR| PG_BINARY,
+ S_IRUSR | S_IWUSR);
+
+ if (fd < 0)
+ ereport(ERROR,
+ (errcode_for_file_access(),
+ errmsg("could not create file \"%s\": %m", fname)));
+
+ if (pg_fsync(fd) != 0)
+ ereport(ERROR,
+ (errcode_for_file_access(),
+ errmsg("could not fsync file \"%s\": %m", fname)));
+ close(fd);
+ }
--
1.6.5.12.gd65df24
--
Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-performance
