This patch adds the --sparse-block option. Andrea Righi writes:
In some filesystems, typically optimized for large I/O throughputs (like
IBM GPFS, IBM SAN FS, or distributed filesystems in general) a lot of
lseek() operations can strongly impact on performances. In this cases it
can be helpful to enlarge the block size used to handle sparse files
directly from a command line parameter.
For example, using a sparse write size of 32KB, I've been able to
increase the transfer rate of an order of magnitude copying the output
files of scientific applications from GPFS to GPFS or GPFS to SAN FS.
-Andrea
To use this patch, run these commands for a successful build:
patch -p1 <patches/sparse-block.diff
./configure (optional if already run)
make
based-on: d73762eea3f15f2c56bb3fa9394ad1883c25c949
diff --git a/fileio.c b/fileio.c
--- a/fileio.c
+++ b/fileio.c
@@ -34,6 +34,7 @@
#define ALIGNED_LENGTH(len) ((((len) - 1) | (ALIGN_BOUNDRY-1)) + 1)
extern int sparse_files;
+extern int sparse_files_block_size;
OFF_T preallocated_len = 0;
@@ -147,7 +148,7 @@ int write_file(int f, int use_seek, OFF_T offset, const char *buf, int len)
while (len > 0) {
int r1;
if (sparse_files > 0) {
- int len1 = MIN(len, SPARSE_WRITE_SIZE);
+ int len1 = MIN(len, sparse_files_block_size);
r1 = write_sparse(f, use_seek, offset, buf, len1);
offset += r1;
} else {
diff --git a/options.c b/options.c
--- a/options.c
+++ b/options.c
@@ -75,6 +75,7 @@ int remove_source_files = 0;
int one_file_system = 0;
int protocol_version = PROTOCOL_VERSION;
int sparse_files = 0;
+long sparse_files_block_size = SPARSE_WRITE_SIZE;
int preallocate_files = 0;
int do_compression = 0;
int def_compress_level = NOT_SPECIFIED;
@@ -715,6 +716,7 @@ void usage(enum logcode F)
rprintf(F," --fake-super store/recover privileged attrs using xattrs\n");
#endif
rprintf(F," -S, --sparse turn sequences of nulls into sparse blocks\n");
+ rprintf(F," --sparse-block=SIZE set the block size used to handle sparse files\n");
#ifdef SUPPORT_PREALLOCATION
rprintf(F," --preallocate allocate dest files before writing them\n");
#else
@@ -924,6 +926,7 @@ static struct poptOption long_options[] = {
{"sparse", 'S', POPT_ARG_VAL, &sparse_files, 1, 0, 0 },
{"no-sparse", 0, POPT_ARG_VAL, &sparse_files, 0, 0, 0 },
{"no-S", 0, POPT_ARG_VAL, &sparse_files, 0, 0, 0 },
+ {"sparse-block", 0, POPT_ARG_LONG, &sparse_files_block_size, 0, 0, 0 },
{"preallocate", 0, POPT_ARG_NONE, &preallocate_files, 0, 0, 0},
{"inplace", 0, POPT_ARG_VAL, &inplace, 1, 0, 0 },
{"no-inplace", 0, POPT_ARG_VAL, &inplace, 0, 0, 0 },
@@ -2600,6 +2603,12 @@ void server_options(char **args, int *argc_p)
args[ac++] = arg;
}
+ if (sparse_files_block_size) {
+ if (asprintf(&arg, "--sparse-block=%lu", sparse_files_block_size) < 0)
+ goto oom;
+ args[ac++] = arg;
+ }
+
if (io_timeout) {
if (asprintf(&arg, "--timeout=%d", io_timeout) < 0)
goto oom;
diff --git a/rsync.yo b/rsync.yo
--- a/rsync.yo
+++ b/rsync.yo
@@ -377,6 +377,7 @@ to the detailed description below for a complete description. verb(
--super receiver attempts super-user activities
--fake-super store/recover privileged attrs using xattrs
-S, --sparse turn sequences of nulls into sparse blocks
+ --sparse-block=SIZE set block size used to handle sparse files
--preallocate allocate dest files before writing
-n, --dry-run perform a trial run with no changes made
-W, --whole-file copy files whole (w/o delta-xfer algorithm)
@@ -1299,6 +1300,15 @@ If combined with bf(--sparse), the file will only have sparse blocks (as
opposed to allocated sequences of null bytes) if the kernel version and
filesystem type support creating holes in the allocated data.
+dit(bf(--sparse-block=SIZE)) Change the block size used to handle sparse files
+to SIZE bytes. This option only has an effect if the bf(--sparse) (bf(-S))
+option was also specified. The default block size used by rsync to detect a
+file hole is 1024 bytes; when the receiver writes data to the destination file
+and option bf(--sparse) is used, rsync checks every 1024-bytes chunk to detect
+if they are actually filled with data or not. With certain filesystems,
+optimized to receive data streams for example, enlarging this block size can
+strongly increase performance. The option can be used to tune this block size.
+
dit(bf(-n, --dry-run)) This makes rsync perform a trial run that doesn't
make any changes (and produces mostly the same output as a real run). It
is most commonly used in combination with the bf(-v, --verbose) and/or
diff -Nurp a/rsync.1 b/rsync.1
--- a/rsync.1
+++ b/rsync.1
@@ -453,6 +453,7 @@ to the detailed description below for a
\-\-super receiver attempts super\-user activities
\-\-fake\-super store/recover privileged attrs using xattrs
\-S, \-\-sparse turn sequences of nulls into sparse blocks
+ \-\-sparse\-block=SIZE set block size used to handle sparse files
\-\-preallocate allocate dest files before writing
\-n, \-\-dry\-run perform a trial run with no changes made
\-W, \-\-whole\-file copy files whole (w/o delta\-xfer algorithm)
@@ -1489,6 +1490,16 @@ If combined with \fB\-\-sparse\fP, the f
opposed to allocated sequences of null bytes) if the kernel version and
filesystem type support creating holes in the allocated data.
.IP
+.IP "\fB\-\-sparse\-block=SIZE\fP"
+Change the block size used to handle sparse files
+to SIZE bytes. This option only has an effect if the \fB\-\-sparse\fP (\fB\-S\fP)
+option was also specified. The default block size used by rsync to detect a
+file hole is 1024 bytes; when the receiver writes data to the destination file
+and option \fB\-\-sparse\fP is used, rsync checks every 1024\-bytes chunk to detect
+if they are actually filled with data or not. With certain filesystems,
+optimized to receive data streams for example, enlarging this block size can
+strongly increase performance. The option can be used to tune this block size.
+.IP
.IP "\fB\-n, \-\-dry\-run\fP"
This makes rsync perform a trial run that doesn\(cq\&t
make any changes (and produces mostly the same output as a real run). It