1GIT-PACK-OBJECTS(1) Git Manual GIT-PACK-OBJECTS(1)
2
6 git-pack-objects - Create a packed archive of objects
7
9 git pack-objects [-q | --progress | --all-progress] [--all-progress-implied]
10 [--no-reuse-delta] [--delta-base-offset] [--non-empty]
11 [--local] [--incremental] [--window=<n>] [--depth=<n>]
12 [--revs [--unpacked | --all]] [--keep-pack=<pack-name>]
13 [--stdout [--filter=<filter-spec>] | <base-name>]
14 [--shallow] [--keep-true-parents] [--[no-]sparse] < <object-list>
15
17 Reads list of objects from the standard input, and writes either one or
18 more packed archives with the specified base-name to disk, or a packed
19 archive to the standard output.
20 A packed archive is an efficient way to transfer a set of objects
22 between two repositories as well as an access efficient archival
23 format. In a packed archive, an object is either stored as a compressed
24 whole or as a difference from some other object. The latter is often
25 called a delta.
26 The packed archive format (.pack) is designed to be self-contained so
28 that it can be unpacked without any further information. Therefore,
29 each object that a delta depends upon must be present within the pack.
30 A pack index file (.idx) is generated for fast, random access to the
32 objects in the pack. Placing both the index file (.idx) and the packed
33 archive (.pack) in the pack/ subdirectory of $GIT_OBJECT_DIRECTORY (or
34 any of the directories on $GIT_ALTERNATE_OBJECT_DIRECTORIES) enables
35 Git to read from the pack archive.
36 The git unpack-objects command can read the packed archive and expand
38 the objects contained in the pack into "one-file one-object" format;
39 this is typically done by the smart-pull commands when a pack is
40 created on-the-fly for efficient network transport by their peers.
41
43 base-name
44 Write into pairs of files (.pack and .idx), using <base-name> to
45 determine the name of the created file. When this option is used,
46 the two files in a pair are written in
47 <base-name>-<SHA-1>.{pack,idx} files. <SHA-1> is a hash based on
48 the pack content and is written to the standard output of the
49 command.
50 --stdout
52 Write the pack contents (what would have been written to .pack
53 file) out to the standard output.
54 --revs
56 Read the revision arguments from the standard input, instead of
57 individual object names. The revision arguments are processed the
58 same way as git rev-list with the --objects flag uses its commit
59 arguments to build the list of objects it outputs. The objects on
60 the resulting list are packed. Besides revisions, --not or
61 --shallow <SHA-1> lines are also accepted.
62 --unpacked
64 This implies --revs. When processing the list of revision arguments
65 read from the standard input, limit the objects packed to those
66 that are not already packed.
67 --all
69 This implies --revs. In addition to the list of revision arguments
70 read from the standard input, pretend as if all refs under refs/
71 are specified to be included.
72 --include-tag
74 Include unasked-for annotated tags if the object they reference was
75 included in the resulting packfile. This can be useful to send new
76 tags to native Git clients.
77 --stdin-packs
79 Read the basenames of packfiles (e.g., pack-1234abcd.pack) from the
80 standard input, instead of object names or revision arguments. The
81 resulting pack contains all objects listed in the included packs
82 (those not beginning with ^), excluding any objects listed in the
83 excluded packs (beginning with ^).
84 Incompatible with --revs, or options that imply --revs (such as
86 --all), with the exception of --unpacked, which is compatible.
87 --window=<n>, --depth=<n>
89 These two options affect how the objects contained in the pack are
90 stored using delta compression. The objects are first internally
91 sorted by type, size and optionally names and compared against the
92 other objects within --window to see if using delta compression
93 saves space. --depth limits the maximum delta depth; making it too
94 deep affects the performance on the unpacker side, because delta
95 data needs to be applied that many times to get to the necessary
96 object.
97 The default value for --window is 10 and --depth is 50. The maximum
99 depth is 4095.
100 --window-memory=<n>
102 This option provides an additional limit on top of --window; the
103 window size will dynamically scale down so as to not take up more
104 than <n> bytes in memory. This is useful in repositories with a mix
105 of large and small objects to not run out of memory with a large
106 window, but still be able to take advantage of the large window for
107 the smaller objects. The size can be suffixed with "k", "m", or
108 "g". --window-memory=0 makes memory usage unlimited. The default
109 is taken from the pack.windowMemory configuration variable.
110 --max-pack-size=<n>
112 In unusual scenarios, you may not be able to create files larger
113 than a certain size on your filesystem, and this option can be used
114 to tell the command to split the output packfile into multiple
115 independent packfiles, each not larger than the given size. The
116 size can be suffixed with "k", "m", or "g". The minimum size
117 allowed is limited to 1 MiB. The default is unlimited, unless the
118 config variable pack.packSizeLimit is set. Note that this option
119 may result in a larger and slower repository; see the discussion in
120 pack.packSizeLimit.
121 --honor-pack-keep
123 This flag causes an object already in a local pack that has a .keep
124 file to be ignored, even if it would have otherwise been packed.
125 --keep-pack=<pack-name>
127 This flag causes an object already in the given pack to be ignored,
128 even if it would have otherwise been packed. <pack-name> is the
129 pack file name without leading directory (e.g. pack-123.pack). The
130 option could be specified multiple times to keep multiple packs.
131 --incremental
133 This flag causes an object already in a pack to be ignored even if
134 it would have otherwise been packed.
135 --local
137 This flag causes an object that is borrowed from an alternate
138 object store to be ignored even if it would have otherwise been
139 packed.
140 --non-empty
142 Only create a packed archive if it would contain at least one
143 object.
144 --progress
146 Progress status is reported on the standard error stream by default
147 when it is attached to a terminal, unless -q is specified. This
148 flag forces progress status even if the standard error stream is
149 not directed to a terminal.
150 --all-progress
152 When --stdout is specified then progress report is displayed during
153 the object count and compression phases but inhibited during the
154 write-out phase. The reason is that in some cases the output stream
155 is directly linked to another command which may wish to display
156 progress status of its own as it processes incoming pack data. This
157 flag is like --progress except that it forces progress report for
158 the write-out phase as well even if --stdout is used.
159 --all-progress-implied
161 This is used to imply --all-progress whenever progress display is
162 activated. Unlike --all-progress this flag doesn’t actually force
163 any progress display by itself.
164 -q
166 This flag makes the command not to report its progress on the
167 standard error stream.
168 --no-reuse-delta
170 When creating a packed archive in a repository that has existing
171 packs, the command reuses existing deltas. This sometimes results
172 in a slightly suboptimal pack. This flag tells the command not to
173 reuse existing deltas but compute them from scratch.
174 --no-reuse-object
176 This flag tells the command not to reuse existing object data at
177 all, including non deltified object, forcing recompression of
178 everything. This implies --no-reuse-delta. Useful only in the
179 obscure case where wholesale enforcement of a different compression
180 level on the packed data is desired.
181 --compression=<n>
183 Specifies compression level for newly-compressed data in the
184 generated pack. If not specified, pack compression level is
185 determined first by pack.compression, then by core.compression, and
186 defaults to -1, the zlib default, if neither is set. Add
187 --no-reuse-object if you want to force a uniform compression level
188 on all data no matter the source.
189 --[no-]sparse
191 Toggle the "sparse" algorithm to determine which objects to include
192 in the pack, when combined with the "--revs" option. This algorithm
193 only walks trees that appear in paths that introduce new objects.
194 This can have significant performance benefits when computing a
195 pack to send a small change. However, it is possible that extra
196 objects are added to the pack-file if the included commits contain
197 certain types of direct renames. If this option is not included, it
198 defaults to the value of pack.useSparse, which is true unless
199 otherwise specified.
200 --thin
202 Create a "thin" pack by omitting the common objects between a
203 sender and a receiver in order to reduce network transfer. This
204 option only makes sense in conjunction with --stdout.
205 Note: A thin pack violates the packed archive format by omitting
207 required objects and is thus unusable by Git without making it
208 self-contained. Use git index-pack --fix-thin (see git-index-
209 pack(1)) to restore the self-contained property.
210 --shallow
212 Optimize a pack that will be provided to a client with a shallow
213 repository. This option, combined with --thin, can result in a
214 smaller pack at the cost of speed.
215 --delta-base-offset
217 A packed archive can express the base object of a delta as either a
218 20-byte object name or as an offset in the stream, but ancient
219 versions of Git don’t understand the latter. By default, git
220 pack-objects only uses the former format for better compatibility.
221 This option allows the command to use the latter format for
222 compactness. Depending on the average delta chain length, this
223 option typically shrinks the resulting packfile by 3-5 per-cent.
224 Note: Porcelain commands such as git gc (see git-gc(1)), git repack
226 (see git-repack(1)) pass this option by default in modern Git when
227 they put objects in your repository into pack files. So does git
228 bundle (see git-bundle(1)) when it creates a bundle.
229 --threads=<n>
231 Specifies the number of threads to spawn when searching for best
232 delta matches. This requires that pack-objects be compiled with
233 pthreads otherwise this option is ignored with a warning. This is
234 meant to reduce packing time on multiprocessor machines. The
235 required amount of memory for the delta search window is however
236 multiplied by the number of threads. Specifying 0 will cause Git to
237 auto-detect the number of CPU’s and set the number of threads
238 accordingly.
239 --index-version=<version>[,<offset>]
241 This is intended to be used by the test suite only. It allows to
242 force the version for the generated pack index, and to force 64-bit
243 index entries on objects located above the given offset.
244 --keep-true-parents
246 With this option, parents that are hidden by grafts are packed
247 nevertheless.
248 --filter=<filter-spec>
250 Requires --stdout. Omits certain objects (usually blobs) from the
251 resulting packfile. See git-rev-list(1) for valid <filter-spec>
252 forms.
253 --no-filter
255 Turns off any previous --filter= argument.
256 --missing=<missing-action>
258 A debug option to help with future "partial clone" development.
259 This option specifies how missing objects are handled.
260 The form --missing=error requests that pack-objects stop with an
262 error if a missing object is encountered. If the repository is a
263 partial clone, an attempt to fetch missing objects will be made
264 before declaring them missing. This is the default action.
265 The form --missing=allow-any will allow object traversal to
267 continue if a missing object is encountered. No fetch of a missing
268 object will occur. Missing objects will silently be omitted from
269 the results.
270 The form --missing=allow-promisor is like allow-any, but will only
272 allow object traversal to continue for EXPECTED promisor missing
273 objects. No fetch of a missing object will occur. An unexpected
274 missing object will raise an error.
275 --exclude-promisor-objects
277 Omit objects that are known to be in the promisor remote. (This
278 option has the purpose of operating only on locally created
279 objects, so that when we repack, we still maintain a distinction
280 between locally created objects [without .promisor] and objects
281 from the promisor remote [with .promisor].) This is used with
282 partial clone.
283 --keep-unreachable
285 Objects unreachable from the refs in packs named with --unpacked=
286 option are added to the resulting pack, in addition to the
287 reachable objects that are not in packs marked with *.keep files.
288 This implies --revs.
289 --pack-loose-unreachable
291 Pack unreachable loose objects (and their loose counterparts
292 removed). This implies --revs.
293 --unpack-unreachable
295 Keep unreachable objects in loose form. This implies --revs.
296 --delta-islands
298 Restrict delta matches based on "islands". See DELTA ISLANDS below.
299
301 When possible, pack-objects tries to reuse existing on-disk deltas to
302 avoid having to search for new ones on the fly. This is an important
303 optimization for serving fetches, because it means the server can avoid
304 inflating most objects at all and just send the bytes directly from
305 disk. This optimization can’t work when an object is stored as a delta
306 against a base which the receiver does not have (and which we are not
307 already sending). In that case the server "breaks" the delta and has to
308 find a new one, which has a high CPU cost. Therefore it’s important for
309 performance that the set of objects in on-disk delta relationships
310 match what a client would fetch.
311 In a normal repository, this tends to work automatically. The objects
313 are mostly reachable from the branches and tags, and that’s what
314 clients fetch. Any deltas we find on the server are likely to be
315 between objects the client has or will have.
316 But in some repository setups, you may have several related but
318 separate groups of ref tips, with clients tending to fetch those groups
319 independently. For example, imagine that you are hosting several
320 "forks" of a repository in a single shared object store, and letting
321 clients view them as separate repositories through GIT_NAMESPACE or
322 separate repos using the alternates mechanism. A naive repack may find
323 that the optimal delta for an object is against a base that is only
324 found in another fork. But when a client fetches, they will not have
325 the base object, and we’ll have to find a new delta on the fly.
326 A similar situation may exist if you have many refs outside of
328 refs/heads/ and refs/tags/ that point to related objects (e.g.,
329 refs/pull or refs/changes used by some hosting providers). By default,
330 clients fetch only heads and tags, and deltas against objects found
331 only in those other groups cannot be sent as-is.
332 Delta islands solve this problem by allowing you to group your refs
334 into distinct "islands". Pack-objects computes which objects are
335 reachable from which islands, and refuses to make a delta from an
336 object A against a base which is not present in all of A's islands.
337 This results in slightly larger packs (because we miss some delta
338 opportunities), but guarantees that a fetch of one island will not have
339 to recompute deltas on the fly due to crossing island boundaries.
340 When repacking with delta islands the delta window tends to get clogged
342 with candidates that are forbidden by the config. Repacking with a big
343 --window helps (and doesn’t take as long as it otherwise might because
344 we can reject some object pairs based on islands before doing any
345 computation on the content).
346 Islands are configured via the pack.island option, which can be
348 specified multiple times. Each value is a left-anchored regular
349 expressions matching refnames. For example:
350 [pack]
352 island = refs/heads/
353 island = refs/tags/
354 puts heads and tags into an island (whose name is the empty string; see
356 below for more on naming). Any refs which do not match those regular
357 expressions (e.g., refs/pull/123) is not in any island. Any object
358 which is reachable only from refs/pull/ (but not heads or tags) is
359 therefore not a candidate to be used as a base for refs/heads/.
360 Refs are grouped into islands based on their "names", and two regexes
362 that produce the same name are considered to be in the same island. The
363 names are computed from the regexes by concatenating any capture groups
364 from the regex, with a - dash in between. (And if there are no capture
365 groups, then the name is the empty string, as in the above example.)
366 This allows you to create arbitrary numbers of islands. Only up to 14
367 such capture groups are supported though.
368 For example, imagine you store the refs for each fork in
370 refs/virtual/ID, where ID is a numeric identifier. You might then
371 configure:
372 [pack]
374 island = refs/virtual/([0-9]+)/heads/
375 island = refs/virtual/([0-9]+)/tags/
376 island = refs/virtual/([0-9]+)/(pull)/
377 That puts the heads and tags for each fork in their own island (named
379 "1234" or similar), and the pull refs for each go into their own
380 "1234-pull".
381 Note that we pick a single island for each regex to go into, using
383 "last one wins" ordering (which allows repo-specific config to take
384 precedence over user-wide config, and so forth).
385
387 Various configuration variables affect packing, see git-config(1)
388 (search for "pack" and "delta").
389 Notably, delta compression is not used on objects larger than the
391 core.bigFileThreshold configuration variable and on files with the
392 attribute delta set to false.
393
395 git-rev-list(1) git-repack(1) git-prune-packed(1)
396
398 Part of the git(1) suite
399Git 2.36.1 2022-05-05 GIT-PACK-OBJECTS(1)