Git fork
1git-maintenance(1)
2==================
3
4NAME
5----
6git-maintenance - Run tasks to optimize Git repository data
7
8
9SYNOPSIS
10--------
11[verse]
12'git maintenance' run [<options>]
13'git maintenance' start [--scheduler=<scheduler>]
14'git maintenance' (stop|register|unregister) [<options>]
15
16
17DESCRIPTION
18-----------
19Run tasks to optimize Git repository data, speeding up other Git commands
20and reducing storage requirements for the repository.
21
22Git commands that add repository data, such as `git add` or `git fetch`,
23are optimized for a responsive user experience. These commands do not take
24time to optimize the Git data, since such optimizations scale with the full
25size of the repository while these user commands each perform a relatively
26small action.
27
28The `git maintenance` command provides flexibility for how to optimize the
29Git repository.
30
31SUBCOMMANDS
32-----------
33
34run::
35 Run one or more maintenance tasks. If one or more `--task` options
36 are specified, then those tasks are run in that order. Otherwise,
37 the tasks are determined by which `maintenance.<task>.enabled`
38 config options are true. By default, only `maintenance.gc.enabled`
39 is true.
40
41start::
42 Start running maintenance on the current repository. This performs
43 the same config updates as the `register` subcommand, then updates
44 the background scheduler to run `git maintenance run --scheduled`
45 on an hourly basis.
46
47stop::
48 Halt the background maintenance schedule. The current repository
49 is not removed from the list of maintained repositories, in case
50 the background maintenance is restarted later.
51
52register::
53 Initialize Git config values so any scheduled maintenance will start
54 running on this repository. This adds the repository to the
55 `maintenance.repo` config variable in the current user's global config,
56 or the config specified by --config-file option, and enables some
57 recommended configuration values for `maintenance.<task>.schedule`. The
58 tasks that are enabled are safe for running in the background without
59 disrupting foreground processes.
60+
61The `register` subcommand will also set the `maintenance.strategy` config
62value to `incremental`, if this value is not previously set. The
63`incremental` strategy uses the following schedule for each maintenance
64task:
65+
66--
67* `gc`: disabled.
68* `commit-graph`: hourly.
69* `prefetch`: hourly.
70* `loose-objects`: daily.
71* `incremental-repack`: daily.
72--
73+
74`git maintenance register` will also disable foreground maintenance by
75setting `maintenance.auto = false` in the current repository. This config
76setting will remain after a `git maintenance unregister` command.
77
78unregister::
79 Remove the current repository from background maintenance. This
80 only removes the repository from the configured list. It does not
81 stop the background maintenance processes from running.
82+
83The `unregister` subcommand will report an error if the current repository
84is not already registered. Use the `--force` option to return success even
85when the current repository is not registered.
86
87TASKS
88-----
89
90commit-graph::
91 The `commit-graph` job updates the `commit-graph` files incrementally,
92 then verifies that the written data is correct. The incremental
93 write is safe to run alongside concurrent Git processes since it
94 will not expire `.graph` files that were in the previous
95 `commit-graph-chain` file. They will be deleted by a later run based
96 on the expiration delay.
97
98prefetch::
99 The `prefetch` task updates the object directory with the latest
100 objects from all registered remotes. For each remote, a `git fetch`
101 command is run. The configured refspec is modified to place all
102 requested refs within `refs/prefetch/`. Also, tags are not updated.
103+
104This is done to avoid disrupting the remote-tracking branches. The end users
105expect these refs to stay unmoved unless they initiate a fetch. However,
106with the prefetch task, the objects necessary to complete a later real fetch
107would already be obtained, making the real fetch faster. In the ideal case,
108it will just become an update to a bunch of remote-tracking branches without
109any object transfer.
110+
111The `remote.<name>.skipFetchAll` configuration can be used to
112exclude a particular remote from getting prefetched.
113
114gc::
115 Clean up unnecessary files and optimize the local repository. "GC"
116 stands for "garbage collection," but this task performs many
117 smaller tasks. This task can be expensive for large repositories,
118 as it repacks all Git objects into a single pack-file. It can also
119 be disruptive in some situations, as it deletes stale data. See
120 linkgit:git-gc[1] for more details on garbage collection in Git.
121
122loose-objects::
123 The `loose-objects` job cleans up loose objects and places them into
124 pack-files. In order to prevent race conditions with concurrent Git
125 commands, it follows a two-step process. First, it deletes any loose
126 objects that already exist in a pack-file; concurrent Git processes
127 will examine the pack-file for the object data instead of the loose
128 object. Second, it creates a new pack-file (starting with "loose-")
129 containing a batch of loose objects.
130+
131The batch size defaults to fifty thousand objects to prevent the job from
132taking too long on a repository with many loose objects. Use the
133`maintenance.loose-objects.batchSize` config option to adjust this size,
134including a value of `0` to remove the limit.
135+
136The `gc` task writes unreachable objects as loose objects to be cleaned up
137by a later step only if they are not re-added to a pack-file; for this
138reason it is not advisable to enable both the `loose-objects` and `gc`
139tasks at the same time.
140
141incremental-repack::
142 The `incremental-repack` job repacks the object directory
143 using the `multi-pack-index` feature. In order to prevent race
144 conditions with concurrent Git commands, it follows a two-step
145 process. First, it calls `git multi-pack-index expire` to delete
146 pack-files unreferenced by the `multi-pack-index` file. Second, it
147 calls `git multi-pack-index repack` to select several small
148 pack-files and repack them into a bigger one, and then update the
149 `multi-pack-index` entries that refer to the small pack-files to
150 refer to the new pack-file. This prepares those small pack-files
151 for deletion upon the next run of `git multi-pack-index expire`.
152 The selection of the small pack-files is such that the expected
153 size of the big pack-file is at least the batch size; see the
154 `--batch-size` option for the `repack` subcommand in
155 linkgit:git-multi-pack-index[1]. The default batch-size is zero,
156 which is a special case that attempts to repack all pack-files
157 into a single pack-file.
158
159pack-refs::
160 The `pack-refs` task collects the loose reference files and
161 collects them into a single file. This speeds up operations that
162 need to iterate across many references. See linkgit:git-pack-refs[1]
163 for more information.
164
165reflog-expire::
166 The `reflog-expire` task deletes any entries in the reflog older than the
167 expiry threshold. See linkgit:git-reflog[1] for more information.
168
169rerere-gc::
170 The `rerere-gc` task invokes garbage collection for stale entries in
171 the rerere cache. See linkgit:git-rerere[1] for more information.
172
173worktree-prune::
174 The `worktree-prune` task deletes stale or broken worktrees. See
175 linkgit:git-worktree[1] for more information.
176
177OPTIONS
178-------
179--auto::
180 When combined with the `run` subcommand, run maintenance tasks
181 only if certain thresholds are met. For example, the `gc` task
182 runs when the number of loose objects exceeds the number stored
183 in the `gc.auto` config setting, or when the number of pack-files
184 exceeds the `gc.autoPackLimit` config setting. Not compatible with
185 the `--schedule` option.
186
187--schedule::
188 When combined with the `run` subcommand, run maintenance tasks
189 only if certain time conditions are met, as specified by the
190 `maintenance.<task>.schedule` config value for each `<task>`.
191 This config value specifies a number of seconds since the last
192 time that task ran, according to the `maintenance.<task>.lastRun`
193 config value. The tasks that are tested are those provided by
194 the `--task=<task>` option(s) or those with
195 `maintenance.<task>.enabled` set to true.
196
197--quiet::
198 Do not report progress or other information over `stderr`.
199
200--task=<task>::
201 If this option is specified one or more times, then only run the
202 specified tasks in the specified order. If no `--task=<task>`
203 arguments are specified, then only the tasks with
204 `maintenance.<task>.enabled` configured as `true` are considered.
205 See the 'TASKS' section for the list of accepted `<task>` values.
206
207--scheduler=auto|crontab|systemd-timer|launchctl|schtasks::
208 When combined with the `start` subcommand, specify the scheduler
209 for running the hourly, daily and weekly executions of
210 `git maintenance run`.
211 Possible values for `<scheduler>` are `auto`, `crontab`
212 (POSIX), `systemd-timer` (Linux), `launchctl` (macOS), and
213 `schtasks` (Windows). When `auto` is specified, the
214 appropriate platform-specific scheduler is used; on Linux,
215 `systemd-timer` is used if available, otherwise
216 `crontab`. Default is `auto`.
217
218
219TROUBLESHOOTING
220---------------
221The `git maintenance` command is designed to simplify the repository
222maintenance patterns while minimizing user wait time during Git commands.
223A variety of configuration options are available to allow customizing this
224process. The default maintenance options focus on operations that complete
225quickly, even on large repositories.
226
227Users may find some cases where scheduled maintenance tasks do not run as
228frequently as intended. Each `git maintenance run` command takes a lock on
229the repository's object database, and this prevents other concurrent
230`git maintenance run` commands from running on the same repository. Without
231this safeguard, competing processes could leave the repository in an
232unpredictable state.
233
234The background maintenance schedule runs `git maintenance run` processes
235on an hourly basis. Each run executes the "hourly" tasks. At midnight,
236that process also executes the "daily" tasks. At midnight on the first day
237of the week, that process also executes the "weekly" tasks. A single
238process iterates over each registered repository, performing the scheduled
239tasks for that frequency. The processes are scheduled to a random minute of
240the hour per client to spread out the load that multiple clients might
241generate (e.g. from prefetching). Depending on the number of registered
242repositories and their sizes, this process may take longer than an hour.
243In this case, multiple `git maintenance run` commands may run on the same
244repository at the same time, colliding on the object database lock. This
245results in one of the two tasks not running.
246
247If you find that some maintenance windows are taking longer than one hour
248to complete, then consider reducing the complexity of your maintenance
249tasks. For example, the `gc` task is much slower than the
250`incremental-repack` task. However, this comes at a cost of a slightly
251larger object database. Consider moving more expensive tasks to be run
252less frequently.
253
254Expert users may consider scheduling their own maintenance tasks using a
255different schedule than is available through `git maintenance start` and
256Git configuration options. These users should be aware of the object
257database lock and how concurrent `git maintenance run` commands behave.
258Further, the `git gc` command should not be combined with
259`git maintenance run` commands. `git gc` modifies the object database
260but does not take the lock in the same way as `git maintenance run`. If
261possible, use `git maintenance run --task=gc` instead of `git gc`.
262
263The following sections describe the mechanisms put in place to run
264background maintenance by `git maintenance start` and how to customize
265them.
266
267BACKGROUND MAINTENANCE ON POSIX SYSTEMS
268---------------------------------------
269
270The standard mechanism for scheduling background tasks on POSIX systems
271is cron(8). This tool executes commands based on a given schedule. The
272current list of user-scheduled tasks can be found by running `crontab -l`.
273The schedule written by `git maintenance start` is similar to this:
274
275-----------------------------------------------------------------------
276# BEGIN GIT MAINTENANCE SCHEDULE
277# The following schedule was created by Git
278# Any edits made in this region might be
279# replaced in the future by a Git command.
280
2810 1-23 * * * "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=hourly
2820 0 * * 1-6 "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=daily
2830 0 * * 0 "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=weekly
284
285# END GIT MAINTENANCE SCHEDULE
286-----------------------------------------------------------------------
287
288The comments are used as a region to mark the schedule as written by Git.
289Any modifications within this region will be completely deleted by
290`git maintenance stop` or overwritten by `git maintenance start`.
291
292The `crontab` entry specifies the full path of the `git` executable to
293ensure that the executed `git` command is the same one with which
294`git maintenance start` was issued independent of `PATH`. If the same user
295runs `git maintenance start` with multiple Git executables, then only the
296latest executable is used.
297
298These commands use `git for-each-repo --config=maintenance.repo` to run
299`git maintenance run --schedule=<frequency>` on each repository listed in
300the multi-valued `maintenance.repo` config option. These are typically
301loaded from the user-specific global config. The `git maintenance` process
302then determines which maintenance tasks are configured to run on each
303repository with each `<frequency>` using the `maintenance.<task>.schedule`
304config options. These values are loaded from the global or repository
305config values.
306
307If the config values are insufficient to achieve your desired background
308maintenance schedule, then you can create your own schedule. If you run
309`crontab -e`, then an editor will load with your user-specific `cron`
310schedule. In that editor, you can add your own schedule lines. You could
311start by adapting the default schedule listed earlier, or you could read
312the crontab(5) documentation for advanced scheduling techniques. Please
313do use the full path and `--exec-path` techniques from the default
314schedule to ensure you are executing the correct binaries in your
315schedule.
316
317
318BACKGROUND MAINTENANCE ON LINUX SYSTEMD SYSTEMS
319-----------------------------------------------
320
321While Linux supports `cron`, depending on the distribution, `cron` may
322be an optional package not necessarily installed. On modern Linux
323distributions, systemd timers are superseding it.
324
325If user systemd timers are available, they will be used as a replacement
326of `cron`.
327
328In this case, `git maintenance start` will create user systemd timer units
329and start the timers. The current list of user-scheduled tasks can be found
330by running `systemctl --user list-timers`. The timers written by `git
331maintenance start` are similar to this:
332
333-----------------------------------------------------------------------
334$ systemctl --user list-timers
335NEXT LEFT LAST PASSED UNIT ACTIVATES
336Thu 2021-04-29 19:00:00 CEST 42min left Thu 2021-04-29 18:00:11 CEST 17min ago git-maintenance@hourly.timer git-maintenance@hourly.service
337Fri 2021-04-30 00:00:00 CEST 5h 42min left Thu 2021-04-29 00:00:11 CEST 18h ago git-maintenance@daily.timer git-maintenance@daily.service
338Mon 2021-05-03 00:00:00 CEST 3 days left Mon 2021-04-26 00:00:11 CEST 3 days ago git-maintenance@weekly.timer git-maintenance@weekly.service
339-----------------------------------------------------------------------
340
341One timer is registered for each `--schedule=<frequency>` option.
342
343The definition of the systemd units can be inspected in the following files:
344
345-----------------------------------------------------------------------
346~/.config/systemd/user/git-maintenance@.timer
347~/.config/systemd/user/git-maintenance@.service
348~/.config/systemd/user/timers.target.wants/git-maintenance@hourly.timer
349~/.config/systemd/user/timers.target.wants/git-maintenance@daily.timer
350~/.config/systemd/user/timers.target.wants/git-maintenance@weekly.timer
351-----------------------------------------------------------------------
352
353`git maintenance start` will overwrite these files and start the timer
354again with `systemctl --user`, so any customization should be done by
355creating a drop-in file, i.e. a `.conf` suffixed file in the
356`~/.config/systemd/user/git-maintenance@.service.d` directory.
357
358`git maintenance stop` will stop the user systemd timers and delete
359the above mentioned files.
360
361For more details, see `systemd.timer(5)`.
362
363
364BACKGROUND MAINTENANCE ON MACOS SYSTEMS
365---------------------------------------
366
367While macOS technically supports `cron`, using `crontab -e` requires
368elevated privileges and the executed process does not have a full user
369context. Without a full user context, Git and its credential helpers
370cannot access stored credentials, so some maintenance tasks are not
371functional.
372
373Instead, `git maintenance start` interacts with the `launchctl` tool,
374which is the recommended way to schedule timed jobs in macOS. Scheduling
375maintenance through `git maintenance (start|stop)` requires some
376`launchctl` features available only in macOS 10.11 or later.
377
378Your user-specific scheduled tasks are stored as XML-formatted `.plist`
379files in `~/Library/LaunchAgents/`. You can see the currently-registered
380tasks using the following command:
381
382-----------------------------------------------------------------------
383$ ls ~/Library/LaunchAgents/org.git-scm.git*
384org.git-scm.git.daily.plist
385org.git-scm.git.hourly.plist
386org.git-scm.git.weekly.plist
387-----------------------------------------------------------------------
388
389One task is registered for each `--schedule=<frequency>` option. To
390inspect how the XML format describes each schedule, open one of these
391`.plist` files in an editor and inspect the `<array>` element following
392the `<key>StartCalendarInterval</key>` element.
393
394`git maintenance start` will overwrite these files and register the
395tasks again with `launchctl`, so any customizations should be done by
396creating your own `.plist` files with distinct names. Similarly, the
397`git maintenance stop` command will unregister the tasks with `launchctl`
398and delete the `.plist` files.
399
400To create more advanced customizations to your background tasks, see
401launchctl.plist(5) for more information.
402
403
404BACKGROUND MAINTENANCE ON WINDOWS SYSTEMS
405-----------------------------------------
406
407Windows does not support `cron` and instead has its own system for
408scheduling background tasks. The `git maintenance start` command uses
409the `schtasks` command to submit tasks to this system. You can inspect
410all background tasks using the Task Scheduler application. The tasks
411added by Git have names of the form `Git Maintenance (<frequency>)`.
412The Task Scheduler GUI has ways to inspect these tasks, but you can also
413export the tasks to XML files and view the details there.
414
415Note that since Git is a console application, these background tasks
416create a console window visible to the current user. This can be changed
417manually by selecting the "Run whether user is logged in or not" option
418in Task Scheduler. This change requires a password input, which is why
419`git maintenance start` does not select it by default.
420
421If you want to customize the background tasks, please rename the tasks
422so future calls to `git maintenance (start|stop)` do not overwrite your
423custom tasks.
424
425CONFIGURATION
426-------------
427
428include::includes/cmd-config-section-all.adoc[]
429
430include::config/maintenance.adoc[]
431
432
433GIT
434---
435Part of the linkgit:git[1] suite