[PATCH 0/2] packfile: freshen the mtime of packfile by bump file

[PATCH 0/2] packfile: freshen the mtime of packfile by bump file

[Sun Chao via GitGitGadget]

packfile: freshen the mtime of packfile by bump file

We've talked about the cache reload through earlier patches(
https://lore.kernel.org/git/pull.1043.git.git.1625943685565.gitgitgadget@gmail.com),
and we stopped because no further evidence can tell NFS client will reload
the page caches if the file mtime changed. So our team have done these
experiments:

Step1: prepare git servers which mount the NFS disk and a big repo

We prepared 3 vms named c1, s1 and s2, we also have a NFS server named n1.
s1 and s2 mount the NFS disk from n1 by:

    mount -t nfs -o vers=3,timeo=600,nolock,noatime,lookupcache=postive,\
    actimeo=3 <n1 ip addr>:/repositories /mnt/repositories


We setup git server services on s1 and s2, so we can clone repos from s1 by
git commands. Then we created a repository under /mnt/repositories, and
pushed large files to the repository, so we can find a large .pack file in
the repository with about 1.2 GB size.

Step2: do first git clone from client after drop caches of s1

First we drop the caches from s1 by:

    sync; echo 3 > /proc/sys/vm/drop_caches


Then we run git command in c1 to clone the huge repository we created in
Step1, at the same time we run the two commands in s1:

    tcpdump -nn host <n1 ip addr> -w 1st_command.pcap
    nfsiostat 1 -p /mnt/repositories


try to get the result and check what happends.

Step3: do new git clones without drop caches of s1

After Step2, we called new git clone command in c1 to clone the huge
repository for serveral times, and also run the commands at the same time:

    tcpdump -nn host <n1 ip addr> -w lots_of_command.pcap
    nfsiostat 1 -p /mnt/repositories


Step4: do new git clones with packfile mtime changed

After Step2 and Step3, we try to touch all the ".pack" files from s2, and we
call a new git clone in c1 to download the huge repository again, and run
the two command in s1 at the same time:

    tcpdump -nn host <n1 ip addr> -w mtime_changed_command.pcap
    nfsiostat 1 -p /mnt/repositories


Result:

We got a about 1.4GB big pcap file during Step2 and Step4, we can find lots
of READ request and response after open it with wireshark. And by
'nfsiostat' command we can see the 'ops/s' and 'KB/s' of 'read' in the
output shows a relatively large value for a while.

But we got a 4MB pcap file in Step3, and open it with wireshark, we can only
find GETATTR and FSSTAT requests and response. And we the 'nfsiostat' always
show 0 in 'ops/s' and 'KB/s' of 'read' part in the output.

We have done Step1 to Step4 serveral times, each time the result are same.

So we can make sure the NFS client will reload the page cache if other NFS
client changes the mtime of the large .pack files. And for git servers which
use filesystem like NFS to manage large repositories, reload large files
that only have mtime changed result big NFS server IOPS pressure and that
also makes the git server slow because the IO is the bottleneck when there
are too many client requests for the same big repositries.

And I do think the team who manage the git servers need a configuration
choise which can enhance the mtime of packfile through another file which
should be small enough or even empty. It should be backward compatibility
when it is in default value, but just as metioned by Ævar before, maybe
somepeople what to use it in mixed-version environment, we should warn them
in documents, but such configuration do big help for some team who run some
servers mount the NFS disks.

Sun Chao (2):
  packfile: rename `derive_filename()` to `derive_pack_filename()`
  packfile: freshen the mtime of packfile by bump file

 Documentation/config/core.txt   |  11 +++
 builtin/index-pack.c            |  19 +----
 cache.h                         |   1 +
 config.c                        |   5 ++
 environment.c                   |   1 +
 object-file.c                   |  30 +++++++-
 packfile.c                      |  25 ++++++-
 packfile.h                      |   7 ++
 t/t5326-pack-mtime-bumpfiles.sh | 118 ++++++++++++++++++++++++++++++++
 9 files changed, 198 insertions(+), 19 deletions(-)
 create mode 100755 t/t5326-pack-mtime-bumpfiles.sh


base-commit: 5d213e46bb7b880238ff5ea3914e940a50ae9369
Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-1066%2Fsunchao9%2Fmaster-v1
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-1066/sunchao9/master-v1
Pull-Request: https://github.com/git/git/pull/1066
-- 
gitgitgadget

[PATCH 1/2] packfile: rename `derive_filename()` to `derive_pack_filename()`

[Sun Chao via GitGitGadget]

From: Sun Chao <16657101987@163.com>

In order to allow some function get a new file name from `.pack` file
with a new suffix, move `derive_filename()` in `builtin/index-pack.c`
to `packfile.c` with a new name `derive_pack_filename(), and export
it from `packfile.h`.

Signed-off-by: Sun Chao <16657101987@163.com>
---
 builtin/index-pack.c | 19 +++----------------
 packfile.c           | 13 +++++++++++++
 packfile.h           |  7 +++++++
 3 files changed, 23 insertions(+), 16 deletions(-)

diff --git a/builtin/index-pack.c b/builtin/index-pack.c
index 8336466865c..3c83789ccef 100644
--- a/builtin/index-pack.c
+++ b/builtin/index-pack.c
@@ -1435,19 +1435,6 @@ static void fix_unresolved_deltas(struct hashfile *f)
 	free(sorted_by_pos);
 }
 
-static const char *derive_filename(const char *pack_name, const char *strip,
-				   const char *suffix, struct strbuf *buf)
-{
-	size_t len;
-	if (!strip_suffix(pack_name, strip, &len) || !len ||
-	    pack_name[len - 1] != '.')
-		die(_("packfile name '%s' does not end with '.%s'"),
-		    pack_name, strip);
-	strbuf_add(buf, pack_name, len);
-	strbuf_addstr(buf, suffix);
-	return buf->buf;
-}
-
 static void write_special_file(const char *suffix, const char *msg,
 			       const char *pack_name, const unsigned char *hash,
 			       const char **report)
@@ -1458,7 +1445,7 @@ static void write_special_file(const char *suffix, const char *msg,
 	int msg_len = strlen(msg);
 
 	if (pack_name)
-		filename = derive_filename(pack_name, "pack", suffix, &name_buf);
+		filename = derive_pack_filename(pack_name, "pack", suffix, &name_buf);
 	else
 		filename = odb_pack_name(&name_buf, hash, suffix);
 
@@ -1853,13 +1840,13 @@ int cmd_index_pack(int argc, const char **argv, const char *prefix)
 	if (from_stdin && hash_algo)
 		die(_("--object-format cannot be used with --stdin"));
 	if (!index_name && pack_name)
-		index_name = derive_filename(pack_name, "pack", "idx", &index_name_buf);
+		index_name = derive_pack_filename(pack_name, "pack", "idx", &index_name_buf);
 
 	opts.flags &= ~(WRITE_REV | WRITE_REV_VERIFY);
 	if (rev_index) {
 		opts.flags |= verify ? WRITE_REV_VERIFY : WRITE_REV;
 		if (index_name)
-			rev_index_name = derive_filename(index_name,
+			rev_index_name = derive_pack_filename(index_name,
 							 "idx", "rev",
 							 &rev_index_name_buf);
 	}
diff --git a/packfile.c b/packfile.c
index 9ef6d982928..315c3da259a 100644
--- a/packfile.c
+++ b/packfile.c
@@ -40,6 +40,19 @@ char *sha1_pack_index_name(const unsigned char *sha1)
 	return odb_pack_name(&buf, sha1, "idx");
 }
 
+const char *derive_pack_filename(const char *pack_name, const char *strip,
+				const char *suffix, struct strbuf *buf)
+{
+	size_t len;
+	if (!strip_suffix(pack_name, strip, &len) || !len ||
+	    pack_name[len - 1] != '.')
+		die(_("packfile name '%s' does not end with '.%s'"),
+		    pack_name, strip);
+	strbuf_add(buf, pack_name, len);
+	strbuf_addstr(buf, suffix);
+	return buf->buf;
+}
+
 static unsigned int pack_used_ctr;
 static unsigned int pack_mmap_calls;
 static unsigned int peak_pack_open_windows;
diff --git a/packfile.h b/packfile.h
index 3ae117a8aef..ff702b22e6a 100644
--- a/packfile.h
+++ b/packfile.h
@@ -31,6 +31,13 @@ char *sha1_pack_name(const unsigned char *sha1);
  */
 char *sha1_pack_index_name(const unsigned char *sha1);
 
+/*
+ * Return the corresponding filename with given suffix from "file_name"
+ * which must has "strip" suffix.
+ */
+const char *derive_pack_filename(const char *file_name, const char *strip,
+		const char *suffix, struct strbuf *buf);
+
 /*
  * Return the basename of the packfile, omitting any containing directory
  * (e.g., "pack-1234abcd[...].pack").
-- 
gitgitgadget

[PATCH 2/2] packfile: freshen the mtime of packfile by bump file

[Sun Chao via GitGitGadget]

From: Sun Chao <16657101987@163.com>

Commit 33d4221c79 (write_sha1_file: freshen existing objects,
2014-10-15) avoid writing existing objects by freshen their
mtime (especially the packfiles contains them) in order to
aid the correct caching, and some process like find_lru_pack
can make good decision. However, this is unfriendly to
incremental backup jobs or services rely on cached file system
when there are large '.pack' files exists.

For example, after packed all objects, use 'write-tree' to
create same commit with the same tree and same environments
such like GIT_COMMITTER_DATE and GIT_AUTHOR_DATE, we can
notice the '.pack' file's mtime changed. Git servers
that use filesystems like NFS will reload the '.pack' files
to file system page cache, which will slow the git commands.

So if we freshen the mtime of packfile by updating a '.bump'
file instead, when we check the mtime of packfile, get it from
'.bump' file also. Large git repository may contains large
'.pack' files, but '.bump' files can be empty. This will avoid
file system page caches reload large files from NFS and then
make git commands faster.

Signed-off-by: Sun Chao <16657101987@163.com>
---
 Documentation/config/core.txt   |  11 +++
 cache.h                         |   1 +
 config.c                        |   5 ++
 environment.c                   |   1 +
 object-file.c                   |  30 +++++++-
 packfile.c                      |  12 +++-
 t/t5326-pack-mtime-bumpfiles.sh | 118 ++++++++++++++++++++++++++++++++
 7 files changed, 175 insertions(+), 3 deletions(-)
 create mode 100755 t/t5326-pack-mtime-bumpfiles.sh

diff --git a/Documentation/config/core.txt b/Documentation/config/core.txt
index c04f62a54a1..963d1b54e7e 100644
--- a/Documentation/config/core.txt
+++ b/Documentation/config/core.txt
@@ -398,6 +398,17 @@ the largest projects.  You probably do not need to adjust this value.
 +
 Common unit suffixes of 'k', 'm', or 'g' are supported.
 
+core.packMtimeToBumpFiles::
+	Normally we avoid writing existing object by freshening the mtime
+	of the *.pack file which contains it in order to aid some processes
+	such like prune. Use a *.bump file instead of *.pack file will
+	avoid file system cache re-sync the large packfiles on filesystems
+	like NFS, and consequently make git commands faster.
++
+The default is 'false' which means the *.pack file will be freshened by
+default. If set to 'true', the file with the '.bump' suffix will be
+created automatically, and it's mtime will be freshened instead.
+
 core.deltaBaseCacheLimit::
 	Maximum number of bytes per thread to reserve for caching base objects
 	that may be referenced by multiple deltified objects.  By storing the
diff --git a/cache.h b/cache.h
index bd4869beee4..a563cbacfa2 100644
--- a/cache.h
+++ b/cache.h
@@ -960,6 +960,7 @@ extern const char *git_hooks_path;
 extern int zlib_compression_level;
 extern int core_compression_level;
 extern int pack_compression_level;
+extern int pack_mtime_to_bumpfiles;
 extern size_t packed_git_window_size;
 extern size_t packed_git_limit;
 extern size_t delta_base_cache_limit;
diff --git a/config.c b/config.c
index f33abeab851..10ccf7c5581 100644
--- a/config.c
+++ b/config.c
@@ -1431,6 +1431,11 @@ static int git_default_core_config(const char *var, const char *value, void *cb)
 		return 0;
 	}
 
+	if (!strcmp(var, "core.packmtimetobumpfiles")) {
+		pack_mtime_to_bumpfiles = git_config_bool(var, value);
+		return 0;
+	}
+
 	if (!strcmp(var, "core.deltabasecachelimit")) {
 		delta_base_cache_limit = git_config_ulong(var, value);
 		return 0;
diff --git a/environment.c b/environment.c
index d6b22ede7ea..5fa26cb3758 100644
--- a/environment.c
+++ b/environment.c
@@ -43,6 +43,7 @@ const char *git_hooks_path;
 int zlib_compression_level = Z_BEST_SPEED;
 int core_compression_level;
 int pack_compression_level = Z_DEFAULT_COMPRESSION;
+int pack_mtime_to_bumpfiles;
 int fsync_object_files;
 size_t packed_git_window_size = DEFAULT_PACKED_GIT_WINDOW_SIZE;
 size_t packed_git_limit = DEFAULT_PACKED_GIT_LIMIT;
diff --git a/object-file.c b/object-file.c
index a8be8994814..434073c17f1 100644
--- a/object-file.c
+++ b/object-file.c
@@ -1994,12 +1994,38 @@ static int freshen_loose_object(const struct object_id *oid)
 static int freshen_packed_object(const struct object_id *oid)
 {
 	struct pack_entry e;
+	struct stat st;
+	struct strbuf name_buf = STRBUF_INIT;
+	const char *filename;
+
 	if (!find_pack_entry(the_repository, oid, &e))
 		return 0;
 	if (e.p->freshened)
 		return 1;
-	if (!freshen_file(e.p->pack_name))
-		return 0;
+
+	filename = e.p->pack_name;
+	if (!pack_mtime_to_bumpfiles) {
+		if (!freshen_file(filename))
+			return 0;
+		e.p->freshened = 1;
+		return 1;
+	}
+
+	filename = derive_pack_filename(filename, "pack", "bump", &name_buf);
+	if (lstat(filename, &st) < 0) {
+		int fd = open(filename, O_CREAT|O_EXCL|O_WRONLY, 0664);
+		if (fd < 0) {
+			// here we need to check it again because other git process may created it
+			if (lstat(filename, &st) < 0)
+				die_errno("unable to create '%s'", filename);
+		} else {
+			close(fd);
+		}
+	} else {
+		if (!freshen_file(filename))
+			return 0;
+	}
+
 	e.p->freshened = 1;
 	return 1;
 }
diff --git a/packfile.c b/packfile.c
index 315c3da259a..f5cee440601 100644
--- a/packfile.c
+++ b/packfile.c
@@ -374,7 +374,7 @@ void close_object_store(struct raw_object_store *o)
 
 void unlink_pack_path(const char *pack_name, int force_delete)
 {
-	static const char *exts[] = {".pack", ".idx", ".rev", ".keep", ".bitmap", ".promisor"};
+	static const char *exts[] = {".pack", ".idx", ".rev", ".keep", ".bitmap", ".promisor", ".bump"};
 	int i;
 	struct strbuf buf = STRBUF_INIT;
 	size_t plen;
@@ -741,6 +741,16 @@ struct packed_git *add_packed_git(const char *path, size_t path_len, int local)
 	p->pack_size = st.st_size;
 	p->pack_local = local;
 	p->mtime = st.st_mtime;
+
+	if (pack_mtime_to_bumpfiles) {
+		struct strbuf name_buf = STRBUF_INIT;
+		const char *filename;
+
+		filename = derive_pack_filename(path, "idx", "bump", &name_buf);
+		if (!stat(filename, &st)) {
+			p->mtime = st.st_mtime;
+		}
+	}
 	if (path_len < the_hash_algo->hexsz ||
 	    get_sha1_hex(path + path_len - the_hash_algo->hexsz, p->hash))
 		hashclr(p->hash);
diff --git a/t/t5326-pack-mtime-bumpfiles.sh b/t/t5326-pack-mtime-bumpfiles.sh
new file mode 100755
index 00000000000..d6d9e6dc446
--- /dev/null
+++ b/t/t5326-pack-mtime-bumpfiles.sh
@@ -0,0 +1,118 @@
+#!/bin/sh
+
+test_description='packfile mtime use bump files'
+. ./test-lib.sh
+
+if stat -c %Y . >/dev/null 2>&1; then
+    get_modified_time() { stat -c %Y "$1" 2>/dev/null; }
+elif stat -f %m . >/dev/null 2>&1; then
+    get_modified_time() { stat -f %m "$1" 2>/dev/null; }
+elif date -r . +%s >/dev/null 2>&1; then
+    get_modified_time() { date -r "$1" +%s 2>/dev/null; }
+else
+    echo 'get_modified_time() is unsupported' >&2
+    get_modified_time() { printf '%s' 0; }
+fi
+
+test_expect_success 'freshen existing packfile without core.packMtimeToBumpFiles' '
+	obj1=$(echo one | git hash-object -w --stdin) &&
+	obj2=$(echo two | git hash-object -w --stdin) &&
+	pack1=$(echo $obj1 | git pack-objects .git/objects/pack/pack) &&
+	pack2=$(echo $obj2 | git pack-objects .git/objects/pack/pack) &&
+	test-tool chmtime =-60 .git/objects/pack/pack-$pack1.* &&
+	test-tool chmtime =-60 .git/objects/pack/pack-$pack2.* &&
+	pack1_mtime=$(get_modified_time .git/objects/pack/pack-$pack1.pack) &&
+	pack2_mtime=$(get_modified_time .git/objects/pack/pack-$pack2.pack) &&
+	(echo one | git hash-object -w --stdin) &&
+	! test_path_exists .git/objects/pack/pack-$pack1.bump &&
+	! test_path_exists .git/objects/pack/pack-$pack2.bump &&
+	pack1_mtime_new=$(get_modified_time .git/objects/pack/pack-$pack1.pack) &&
+	pack2_mtime_new=$(get_modified_time .git/objects/pack/pack-$pack2.pack) &&
+	echo "$pack1_mtime : $pack1_mtime_new" &&
+	test ! "$pack1_mtime" = "$pack1_mtime_new" &&
+	test "$pack2_mtime" = "$pack2_mtime_new"
+
+'
+
+test_expect_success 'freshen existing packfile with core.packMtimeToBumpFiles' '
+
+	rm -rf .git/objects && git init &&
+	obj1=$(echo one | git hash-object -w --stdin) &&
+	obj2=$(echo two | git hash-object -w --stdin) &&
+	pack1=$(echo $obj1 | git pack-objects .git/objects/pack/pack) &&
+	pack2=$(echo $obj2 | git pack-objects .git/objects/pack/pack) &&
+	test-tool chmtime =-60 .git/objects/pack/pack-$pack1.* &&
+	test-tool chmtime =-60 .git/objects/pack/pack-$pack2.* &&
+	pack1_mtime=$(get_modified_time .git/objects/pack/pack-$pack1.pack) &&
+	pack2_mtime=$(get_modified_time .git/objects/pack/pack-$pack2.pack) &&
+	(echo one | git -c core.packMtimeToBumpFiles=true hash-object -w --stdin) &&
+	test_path_exists .git/objects/pack/pack-$pack1.bump &&
+	! test_path_exists .git/objects/pack/pack-$pack2.bump &&
+	pack1_mtime_new=$(get_modified_time .git/objects/pack/pack-$pack1.pack) &&
+	pack2_mtime_new=$(get_modified_time .git/objects/pack/pack-$pack2.pack) &&
+	test "$pack1_mtime" = "$pack1_mtime_new" &&
+	test "$pack2_mtime" = "$pack2_mtime_new"
+
+'
+
+test_expect_success 'repack prune unreachable objects without core.packMtimeToBumpFiles' '
+
+	rm -rf .git/objects && git init &&
+	obj1=$(echo one | git hash-object -w --stdin) &&
+	obj2=$(echo two | git hash-object -w --stdin) &&
+	pack1=$(echo $obj1 | git pack-objects .git/objects/pack/pack) &&
+	pack2=$(echo $obj2 | git pack-objects .git/objects/pack/pack) &&
+	echo one | git hash-object -w --stdin &&
+	echo two | git hash-object -w --stdin &&
+	! test_path_exists .git/objects/pack/pack-$pack1.bump &&
+	! test_path_exists .git/objects/pack/pack-$pack2.bump &&
+	git prune-packed &&
+	git cat-file -p $obj1 &&
+	git cat-file -p $obj2 &&
+	test-tool chmtime =-86400 .git/objects/pack/pack-$pack2.pack &&
+	git repack -A -d --unpack-unreachable=1.hour.ago &&
+	git cat-file -p $obj1 &&
+	test_must_fail git cat-file -p $obj2
+
+'
+
+test_expect_success 'repack prune unreachable objects with core.packMtimeToBumpFiles and bump files' '
+
+	rm -rf .git/objects && git init &&
+	obj1=$(echo one | git hash-object -w --stdin) &&
+	obj2=$(echo two | git hash-object -w --stdin) &&
+	pack1=$(echo $obj1 | git pack-objects .git/objects/pack/pack) &&
+	pack2=$(echo $obj2 | git pack-objects .git/objects/pack/pack) &&
+	echo one | git -c core.packMtimeToBumpFiles=true hash-object -w --stdin &&
+	echo two | git -c core.packMtimeToBumpFiles=true hash-object -w --stdin &&
+	test_path_exists .git/objects/pack/pack-$pack1.bump &&
+	test_path_exists .git/objects/pack/pack-$pack2.bump &&
+	test-tool chmtime =-86400 .git/objects/pack/pack-$pack2.pack &&
+	git -c core.packMtimeToBumpFiles=true repack -A -d --unpack-unreachable=1.hour.ago &&
+	git cat-file -p $obj1 &&
+	git cat-file -p $obj2
+
+'
+
+test_expect_success 'repack prune unreachable objects with core.packMtimeToBumpFiles and old bump files' '
+
+	rm -rf .git/objects && git init &&
+	obj1=$(echo one | git hash-object -w --stdin) &&
+	obj2=$(echo two | git hash-object -w --stdin) &&
+	pack1=$(echo $obj1 | git pack-objects .git/objects/pack/pack) &&
+	pack2=$(echo $obj2 | git pack-objects .git/objects/pack/pack) &&
+	echo one | git -c core.packMtimeToBumpFiles=true hash-object -w --stdin &&
+	echo two | git -c core.packMtimeToBumpFiles=true hash-object -w --stdin &&
+	test_path_exists .git/objects/pack/pack-$pack1.bump &&
+	test_path_exists .git/objects/pack/pack-$pack2.bump &&
+	git prune-packed &&
+	git cat-file -p $obj1 &&
+	git cat-file -p $obj2 &&
+	test-tool chmtime =-86400 .git/objects/pack/pack-$pack2.bump &&
+	git -c core.packMtimeToBumpFiles=true repack -A -d --unpack-unreachable=1.hour.ago &&
+	git cat-file -p $obj1 &&
+	test_must_fail git cat-file -p $obj2
+
+'
+
+test_done
-- 
gitgitgadget