1GIT-MAINTENANCE(1) Git Manual GIT-MAINTENANCE(1)
2
3
4
6 git-maintenance - Run tasks to optimize Git repository data
7
9 git maintenance run [<options>]
10 git maintenance start [--scheduler=<scheduler>]
11 git maintenance (stop|register|unregister)
12
14 Run tasks to optimize Git repository data, speeding up other Git
15 commands and reducing storage requirements for the repository.
16
17 Git commands that add repository data, such as git add or git fetch,
18 are optimized for a responsive user experience. These commands do not
19 take time to optimize the Git data, since such optimizations scale with
20 the full size of the repository while these user commands each perform
21 a relatively small action.
22
23 The git maintenance command provides flexibility for how to optimize
24 the Git repository.
25
27 run
28 Run one or more maintenance tasks. If one or more --task options
29 are specified, then those tasks are run in that order. Otherwise,
30 the tasks are determined by which maintenance.<task>.enabled config
31 options are true. By default, only maintenance.gc.enabled is true.
32
33 start
34 Start running maintenance on the current repository. This performs
35 the same config updates as the register subcommand, then updates
36 the background scheduler to run git maintenance run --scheduled on
37 an hourly basis.
38
39 stop
40 Halt the background maintenance schedule. The current repository is
41 not removed from the list of maintained repositories, in case the
42 background maintenance is restarted later.
43
44 register
45 Initialize Git config values so any scheduled maintenance will
46 start running on this repository. This adds the repository to the
47 maintenance.repo config variable in the current user’s global
48 config and enables some recommended configuration values for
49 maintenance.<task>.schedule. The tasks that are enabled are safe
50 for running in the background without disrupting foreground
51 processes.
52
53 The register subcommand will also set the maintenance.strategy
54 config value to incremental, if this value is not previously set.
55 The incremental strategy uses the following schedule for each
56 maintenance task:
57
58 • gc: disabled.
59
60 • commit-graph: hourly.
61
62 • prefetch: hourly.
63
64 • loose-objects: daily.
65
66 • incremental-repack: daily.
67
68 git maintenance register will also disable foreground maintenance
69 by setting maintenance.auto = false in the current repository. This
70 config setting will remain after a git maintenance unregister
71 command.
72
73 unregister
74 Remove the current repository from background maintenance. This
75 only removes the repository from the configured list. It does not
76 stop the background maintenance processes from running.
77
79 commit-graph
80 The commit-graph job updates the commit-graph files incrementally,
81 then verifies that the written data is correct. The incremental
82 write is safe to run alongside concurrent Git processes since it
83 will not expire .graph files that were in the previous
84 commit-graph-chain file. They will be deleted by a later run based
85 on the expiration delay.
86
87 prefetch
88 The prefetch task updates the object directory with the latest
89 objects from all registered remotes. For each remote, a git fetch
90 command is run. The configured refspec is modified to place all
91 requested refs within refs/prefetch/. Also, tags are not updated.
92
93 This is done to avoid disrupting the remote-tracking branches. The
94 end users expect these refs to stay unmoved unless they initiate a
95 fetch. With prefetch task, however, the objects necessary to
96 complete a later real fetch would already be obtained, so the real
97 fetch would go faster. In the ideal case, it will just become an
98 update to a bunch of remote-tracking branches without any object
99 transfer.
100
101 gc
102 Clean up unnecessary files and optimize the local repository. "GC"
103 stands for "garbage collection," but this task performs many
104 smaller tasks. This task can be expensive for large repositories,
105 as it repacks all Git objects into a single pack-file. It can also
106 be disruptive in some situations, as it deletes stale data. See
107 git-gc(1) for more details on garbage collection in Git.
108
109 loose-objects
110 The loose-objects job cleans up loose objects and places them into
111 pack-files. In order to prevent race conditions with concurrent Git
112 commands, it follows a two-step process. First, it deletes any
113 loose objects that already exist in a pack-file; concurrent Git
114 processes will examine the pack-file for the object data instead of
115 the loose object. Second, it creates a new pack-file (starting with
116 "loose-") containing a batch of loose objects. The batch size is
117 limited to 50 thousand objects to prevent the job from taking too
118 long on a repository with many loose objects. The gc task writes
119 unreachable objects as loose objects to be cleaned up by a later
120 step only if they are not re-added to a pack-file; for this reason
121 it is not advisable to enable both the loose-objects and gc tasks
122 at the same time.
123
124 incremental-repack
125 The incremental-repack job repacks the object directory using the
126 multi-pack-index feature. In order to prevent race conditions with
127 concurrent Git commands, it follows a two-step process. First, it
128 calls git multi-pack-index expire to delete pack-files unreferenced
129 by the multi-pack-index file. Second, it calls git multi-pack-index
130 repack to select several small pack-files and repack them into a
131 bigger one, and then update the multi-pack-index entries that refer
132 to the small pack-files to refer to the new pack-file. This
133 prepares those small pack-files for deletion upon the next run of
134 git multi-pack-index expire. The selection of the small pack-files
135 is such that the expected size of the big pack-file is at least the
136 batch size; see the --batch-size option for the repack subcommand
137 in git-multi-pack-index(1). The default batch-size is zero, which
138 is a special case that attempts to repack all pack-files into a
139 single pack-file.
140
141 pack-refs
142 The pack-refs task collects the loose reference files and collects
143 them into a single file. This speeds up operations that need to
144 iterate across many references. See git-pack-refs(1) for more
145 information.
146
148 --auto
149 When combined with the run subcommand, run maintenance tasks only
150 if certain thresholds are met. For example, the gc task runs when
151 the number of loose objects exceeds the number stored in the
152 gc.auto config setting, or when the number of pack-files exceeds
153 the gc.autoPackLimit config setting. Not compatible with the
154 --schedule option.
155
156 --schedule
157 When combined with the run subcommand, run maintenance tasks only
158 if certain time conditions are met, as specified by the
159 maintenance.<task>.schedule config value for each <task>. This
160 config value specifies a number of seconds since the last time that
161 task ran, according to the maintenance.<task>.lastRun config value.
162 The tasks that are tested are those provided by the --task=<task>
163 option(s) or those with maintenance.<task>.enabled set to true.
164
165 --quiet
166 Do not report progress or other information over stderr.
167
168 --task=<task>
169 If this option is specified one or more times, then only run the
170 specified tasks in the specified order. If no --task=<task>
171 arguments are specified, then only the tasks with
172 maintenance.<task>.enabled configured as true are considered. See
173 the TASKS section for the list of accepted <task> values.
174
175 --scheduler=auto|crontab|systemd-timer|launchctl|schtasks
176 When combined with the start subcommand, specify the scheduler for
177 running the hourly, daily and weekly executions of git maintenance
178 run. Possible values for <scheduler> are auto, crontab (POSIX),
179 systemd-timer (Linux), launchctl (macOS), and schtasks (Windows).
180 When auto is specified, the appropriate platform-specific scheduler
181 is used; on Linux, systemd-timer is used if available, otherwise
182 crontab. Default is auto.
183
185 The git maintenance command is designed to simplify the repository
186 maintenance patterns while minimizing user wait time during Git
187 commands. A variety of configuration options are available to allow
188 customizing this process. The default maintenance options focus on
189 operations that complete quickly, even on large repositories.
190
191 Users may find some cases where scheduled maintenance tasks do not run
192 as frequently as intended. Each git maintenance run command takes a
193 lock on the repository’s object database, and this prevents other
194 concurrent git maintenance run commands from running on the same
195 repository. Without this safeguard, competing processes could leave the
196 repository in an unpredictable state.
197
198 The background maintenance schedule runs git maintenance run processes
199 on an hourly basis. Each run executes the "hourly" tasks. At midnight,
200 that process also executes the "daily" tasks. At midnight on the first
201 day of the week, that process also executes the "weekly" tasks. A
202 single process iterates over each registered repository, performing the
203 scheduled tasks for that frequency. Depending on the number of
204 registered repositories and their sizes, this process may take longer
205 than an hour. In this case, multiple git maintenance run commands may
206 run on the same repository at the same time, colliding on the object
207 database lock. This results in one of the two tasks not running.
208
209 If you find that some maintenance windows are taking longer than one
210 hour to complete, then consider reducing the complexity of your
211 maintenance tasks. For example, the gc task is much slower than the
212 incremental-repack task. However, this comes at a cost of a slightly
213 larger object database. Consider moving more expensive tasks to be run
214 less frequently.
215
216 Expert users may consider scheduling their own maintenance tasks using
217 a different schedule than is available through git maintenance start
218 and Git configuration options. These users should be aware of the
219 object database lock and how concurrent git maintenance run commands
220 behave. Further, the git gc command should not be combined with git
221 maintenance run commands. git gc modifies the object database but does
222 not take the lock in the same way as git maintenance run. If possible,
223 use git maintenance run --task=gc instead of git gc.
224
225 The following sections describe the mechanisms put in place to run
226 background maintenance by git maintenance start and how to customize
227 them.
228
230 The standard mechanism for scheduling background tasks on POSIX systems
231 is cron(8). This tool executes commands based on a given schedule. The
232 current list of user-scheduled tasks can be found by running crontab
233 -l. The schedule written by git maintenance start is similar to this:
234
235 # BEGIN GIT MAINTENANCE SCHEDULE
236 # The following schedule was created by Git
237 # Any edits made in this region might be
238 # replaced in the future by a Git command.
239
240 0 1-23 * * * "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=hourly
241 0 0 * * 1-6 "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=daily
242 0 0 * * 0 "/<path>/git" --exec-path="/<path>" for-each-repo --config=maintenance.repo maintenance run --schedule=weekly
243
244 # END GIT MAINTENANCE SCHEDULE
245
246 The comments are used as a region to mark the schedule as written by
247 Git. Any modifications within this region will be completely deleted by
248 git maintenance stop or overwritten by git maintenance start.
249
250 The crontab entry specifies the full path of the git executable to
251 ensure that the executed git command is the same one with which git
252 maintenance start was issued independent of PATH. If the same user runs
253 git maintenance start with multiple Git executables, then only the
254 latest executable is used.
255
256 These commands use git for-each-repo --config=maintenance.repo to run
257 git maintenance run --schedule=<frequency> on each repository listed in
258 the multi-valued maintenance.repo config option. These are typically
259 loaded from the user-specific global config. The git maintenance
260 process then determines which maintenance tasks are configured to run
261 on each repository with each <frequency> using the
262 maintenance.<task>.schedule config options. These values are loaded
263 from the global or repository config values.
264
265 If the config values are insufficient to achieve your desired
266 background maintenance schedule, then you can create your own schedule.
267 If you run crontab -e, then an editor will load with your user-specific
268 cron schedule. In that editor, you can add your own schedule lines. You
269 could start by adapting the default schedule listed earlier, or you
270 could read the crontab(5) documentation for advanced scheduling
271 techniques. Please do use the full path and --exec-path techniques from
272 the default schedule to ensure you are executing the correct binaries
273 in your schedule.
274
276 While Linux supports cron, depending on the distribution, cron may be
277 an optional package not necessarily installed. On modern Linux
278 distributions, systemd timers are superseding it.
279
280 If user systemd timers are available, they will be used as a
281 replacement of cron.
282
283 In this case, git maintenance start will create user systemd timer
284 units and start the timers. The current list of user-scheduled tasks
285 can be found by running systemctl --user list-timers. The timers
286 written by git maintenance start are similar to this:
287
288 $ systemctl --user list-timers
289 NEXT LEFT LAST PASSED UNIT ACTIVATES
290 Thu 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
291 Fri 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
292 Mon 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
293
294 One timer is registered for each --schedule=<frequency> option.
295
296 The definition of the systemd units can be inspected in the following
297 files:
298
299 ~/.config/systemd/user/git-maintenance@.timer
300 ~/.config/systemd/user/git-maintenance@.service
301 ~/.config/systemd/user/timers.target.wants/git-maintenance@hourly.timer
302 ~/.config/systemd/user/timers.target.wants/git-maintenance@daily.timer
303 ~/.config/systemd/user/timers.target.wants/git-maintenance@weekly.timer
304
305 git maintenance start will overwrite these files and start the timer
306 again with systemctl --user, so any customization should be done by
307 creating a drop-in file, i.e. a .conf suffixed file in the
308 ~/.config/systemd/user/git-maintenance@.service.d directory.
309
310 git maintenance stop will stop the user systemd timers and delete the
311 above mentioned files.
312
313 For more details, see systemd.timer(5).
314
316 While macOS technically supports cron, using crontab -e requires
317 elevated privileges and the executed process does not have a full user
318 context. Without a full user context, Git and its credential helpers
319 cannot access stored credentials, so some maintenance tasks are not
320 functional.
321
322 Instead, git maintenance start interacts with the launchctl tool, which
323 is the recommended way to schedule timed jobs in macOS. Scheduling
324 maintenance through git maintenance (start|stop) requires some
325 launchctl features available only in macOS 10.11 or later.
326
327 Your user-specific scheduled tasks are stored as XML-formatted .plist
328 files in ~/Library/LaunchAgents/. You can see the currently-registered
329 tasks using the following command:
330
331 $ ls ~/Library/LaunchAgents/org.git-scm.git*
332 org.git-scm.git.daily.plist
333 org.git-scm.git.hourly.plist
334 org.git-scm.git.weekly.plist
335
336 One task is registered for each --schedule=<frequency> option. To
337 inspect how the XML format describes each schedule, open one of these
338 .plist files in an editor and inspect the <array> element following the
339 <key>StartCalendarInterval</key> element.
340
341 git maintenance start will overwrite these files and register the tasks
342 again with launchctl, so any customizations should be done by creating
343 your own .plist files with distinct names. Similarly, the git
344 maintenance stop command will unregister the tasks with launchctl and
345 delete the .plist files.
346
347 To create more advanced customizations to your background tasks, see
348 launchctl.plist(5) for more information.
349
351 Windows does not support cron and instead has its own system for
352 scheduling background tasks. The git maintenance start command uses the
353 schtasks command to submit tasks to this system. You can inspect all
354 background tasks using the Task Scheduler application. The tasks added
355 by Git have names of the form Git Maintenance (<frequency>). The Task
356 Scheduler GUI has ways to inspect these tasks, but you can also export
357 the tasks to XML files and view the details there.
358
359 Note that since Git is a console application, these background tasks
360 create a console window visible to the current user. This can be
361 changed manually by selecting the "Run whether user is logged in or
362 not" option in Task Scheduler. This change requires a password input,
363 which is why git maintenance start does not select it by default.
364
365 If you want to customize the background tasks, please rename the tasks
366 so future calls to git maintenance (start|stop) do not overwrite your
367 custom tasks.
368
370 Part of the git(1) suite
371
372
373
374Git 2.36.1 2022-05-05 GIT-MAINTENANCE(1)