[PATCH v3] packfile: freshen the mtime of packfile by configuration

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 mount the same NFS disk will re-sync the '.pack' files to cached file system which will slow the git commands.

So if add core.freshenPackfiles to indicate whether or not packs can be freshened, turning off this option on some servers can speed up the execution of some commands on servers which use NFS disk instead of local disk.

Signed-off-by: Sun Chao <16657101987@163.com>
---
    packfile: freshen the mtime of packfile by configuration
    
    packfile: freshen the mtime of packfile by configuration
    
    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 mount the same NFS disk will re-sync the
    '.pack' files to cached file system which will slow the git commands.
    
    Here we can find the description of the cached file system for NFS
    Client from
    https://www.ibm.com/docs/en/aix/7.2?topic=performance-cache-file-system:
    
    3. To ensure that the cached directories and files are kept up to date, 
    CacheFS periodically checks the consistency of files stored in the cache.
    It does this by comparing the current modification time to the previous
    modification time.
    
    4. If the modification times are different, all data and attributes
    for the directory or file are purged from the cache, and new data and
    attributes are retrieved from the back file system.
    
    
    So if add core.freshenPackfiles to indicate whether or not packs can be
    freshened, turning off this option on some servers can speed up the
    execution of some commands on servers which use NFS disk instead of
    local disk.
    
    Signed-off-by: Sun Chao 16657101987@163.com

Published-As: https://github.com/gitgitgadget/git/releases/tag/pr-git-1043%2Fsunchao9%2Fmaster-v3
Fetch-It-Via: git fetch https://github.com/gitgitgadget/git pr-git-1043/sunchao9/master-v3
Pull-Request: https://github.com/git/git/pull/1043

Range-diff vs v2:

 1:  943e31e8587 ! 1:  16c68923bea packfile: freshen the mtime of packfile by configuration
     @@ Commit message
          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 file system
     -    cache when there are large '.pack' files exists.
     +    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, and '.idx' file not.
     +    notice the '.pack' file's mtime changed. Git servers
     +    that mount the same NFS disk will re-sync the '.pack' files
     +    to cached file system which will slow the git commands.
      
     -    If we freshen the mtime of packfile by updating another
     -    file instead of '.pack' file e.g. a empty '.bump' file,
     -    when we need to check the mtime of packfile, get it from
     -    another file instead. Large git repository may contains
     -    large '.pack' files, and we can use smaller files even empty
     -    file to do the mtime get/set operation, this can avoid
     -    file system cache re-sync large '.pack' files again and
     -    then speed up most git commands.
     +    So if add core.freshenPackfiles to indicate whether or not
     +    packs can be freshened, turning off this option on some
     +    servers can speed up the execution of some commands on servers
     +    which use NFS disk instead of local disk.
      
          Signed-off-by: Sun Chao <16657101987@163.com>
      
     @@ Documentation/config/core.txt: the largest projects.  You probably do not need t
       +
       Common unit suffixes of 'k', 'm', or 'g' are supported.
       
     -+core.packMtimeSuffix::
     ++core.freshenPackFiles::
      +	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 different file instead of *.pack file will
     -+	avoid file system cache re-sync the large packfiles, and consequently
     -+	make git commands faster.
     ++	such like prune. Turning off this option on some servers can speed
     ++	up the execution of some commands like 'git-upload-pack'(e.g. some
     ++	servers that mount the same NFS disk will re-sync the *.pack files
     ++	to cached file system if the mtime cahnges).
      ++
     -+The default is 'pack' which means the *.pack file will be freshened by
     -+default. You can configure a different suffix to use, the file with the
     -+suffix will be created automatically, it's better not using any known
     -+suffix such like 'idx', 'keep', 'promisor'.
     ++The default is true which means the *.pack file will be freshened if we
     ++want to write a existing object whthin it.
      +
       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
      
     - ## builtin/index-pack.c ##
     -@@ builtin/index-pack.c: 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)
     -@@ builtin/index-pack.c: 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);
     - 
     -@@ builtin/index-pack.c: 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);
     - 	}
     -
       ## cache.h ##
      @@ cache.h: extern size_t packed_git_limit;
       extern size_t delta_base_cache_limit;
       extern unsigned long big_file_threshold;
       extern unsigned long pack_size_limit_cfg;
     -+extern const char *pack_mtime_suffix;
     ++extern int core_freshen_packfiles;
       
       /*
        * Accessors for the core.sharedrepository config which lazy-load the value
     @@ config.c: static int git_default_core_config(const char *var, const char *value,
       		return 0;
       	}
       
     -+	if (!strcmp(var, "core.packmtimesuffix")) {
     -+		return git_config_string(&pack_mtime_suffix, var, value);
     ++	if (!strcmp(var, "core.freshenpackfiles")) {
     ++		core_freshen_packfiles = git_config_bool(var, value);
      +	}
      +
       	if (!strcmp(var, "core.deltabasecachelimit")) {
     @@ config.c: static int git_default_core_config(const char *var, const char *value,
       		return 0;
      
       ## environment.c ##
     -@@ environment.c: const char *git_hooks_path;
     - int zlib_compression_level = Z_BEST_SPEED;
     - int core_compression_level;
     - int pack_compression_level = Z_DEFAULT_COMPRESSION;
     -+const char *pack_mtime_suffix = "pack";
     - int fsync_object_files;
     - size_t packed_git_window_size = DEFAULT_PACKED_GIT_WINDOW_SIZE;
     - size_t packed_git_limit = DEFAULT_PACKED_GIT_LIMIT;
     +@@ environment.c: int core_sparse_checkout_cone;
     + int merge_log_config = -1;
     + int precomposed_unicode = -1; /* see probe_utf8_pathname_composition() */
     + unsigned long pack_size_limit_cfg;
     ++int core_freshen_packfiles = 1;
     + enum log_refs_config log_all_ref_updates = LOG_REFS_UNSET;
     + 
     + #ifndef PROTECT_HFS_DEFAULT
      
       ## object-file.c ##
      @@ object-file.c: 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 (!core_freshen_packfiles)
     ++		return 1;
      +
       	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 (!strcasecmp(pack_mtime_suffix, "pack")) {
     -+		if (!freshen_file(filename))
     -+			return 0;
     -+		e.p->freshened = 1;
     -+		return 1;
     -+	}
     -+
     -+	/* If we want to freshen different file instead of .pack file, we need
     -+	 * to make sure the file exists and create it if needed.
     -+	 */
     -+	filename = derive_pack_filename(filename, "pack", pack_mtime_suffix, &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;
     - }
     -
     - ## packfile.c ##
     -@@ packfile.c: 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;
     -@@ packfile.c: 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;
     -+
     -+	/* If we have different file used to freshen the mtime, we should
     -+	 * use it at a higher priority.
     -+	 */
     -+	if (!!strcasecmp(pack_mtime_suffix, "pack")) {
     -+		struct strbuf name_buf = STRBUF_INIT;
     -+		const char *filename;
     -+
     -+		filename = derive_pack_filename(path, "idx", pack_mtime_suffix, &name_buf);
     -+		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))
     -
     - ## packfile.h ##
     -@@ packfile.h: 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").
      
       ## t/t7701-repack-unpack-unreachable.sh ##
      @@ t/t7701-repack-unpack-unreachable.sh: test_expect_success 'do not bother loosening old objects' '
       	test_must_fail git cat-file -p $obj2
       '
       
     -+test_expect_success 'do not bother loosening old objects with core.packmtimesuffix config' '
     ++test_expect_success 'do not bother loosening old objects without freshen pack time' '
      +	obj1=$(echo three | git hash-object -w --stdin) &&
      +	obj2=$(echo four | git hash-object -w --stdin) &&
     -+	pack1=$(echo $obj1 | git -c core.packmtimesuffix=bump pack-objects .git/objects/pack/pack) &&
     -+	pack2=$(echo $obj2 | git -c core.packmtimesuffix=bump pack-objects .git/objects/pack/pack) &&
     -+	git -c core.packmtimesuffix=bump prune-packed &&
     ++	pack1=$(echo $obj1 | git -c core.freshenPackFiles=false pack-objects .git/objects/pack/pack) &&
     ++	pack2=$(echo $obj2 | git -c core.freshenPackFiles=false pack-objects .git/objects/pack/pack) &&
     ++	git -c core.freshenPackFiles=false prune-packed &&
      +	git cat-file -p $obj1 &&
      +	git cat-file -p $obj2 &&
     -+	touch .git/objects/pack/pack-$pack2.bump &&
     -+	test-tool chmtime =-86400 .git/objects/pack/pack-$pack2.bump &&
     -+	git -c core.packmtimesuffix=bump repack -A -d --unpack-unreachable=1.hour.ago &&
     ++	test-tool chmtime =-86400 .git/objects/pack/pack-$pack2.pack &&
     ++	git -c core.freshenPackFiles=false repack -A -d --unpack-unreachable=1.hour.ago &&
      +	git cat-file -p $obj1 &&
      +	test_must_fail git cat-file -p $obj2
      +'

 Documentation/config/core.txt        | 11 +++++++++++
 cache.h                              |  1 +
 config.c                             |  4 ++++
 environment.c                        |  1 +
 object-file.c                        |  4 ++++
 t/t7701-repack-unpack-unreachable.sh | 14 ++++++++++++++
 6 files changed, 35 insertions(+)