Documentation/git-maintenance.txt

   1 git-maintenance(1)
   2 ==================
   3
   4 NAME
   5 ----
   6 git-maintenance - Run tasks to optimize Git repository data
   7
   8
   9 SYNOPSIS
  10 --------
  11 [verse]
  12 'git maintenance' run [<options>]
  13
  14
  15 DESCRIPTION
  16 -----------
  17 Run tasks to optimize Git repository data, speeding up other Git commands
  18 and reducing storage requirements for the repository.
  19
  20 Git commands that add repository data, such as `git add` or `git fetch`,
  21 are optimized for a responsive user experience. These commands do not take
  22 time to optimize the Git data, since such optimizations scale with the full
  23 size of the repository while these user commands each perform a relatively
  24 small action.
  25
  26 The `git maintenance` command provides flexibility for how to optimize the
  27 Git repository.
  28
  29 SUBCOMMANDS
  30 -----------
  31
  32 register::
  33         Initialize Git config values so any scheduled maintenance will
  34         start running on this repository. This adds the repository to the
  35         `maintenance.repo` config variable in the current user's global
  36         config and enables some recommended configuration values for
  37         `maintenance.<task>.schedule`. The tasks that are enabled are safe
  38         for running in the background without disrupting foreground
  39         processes.
  40 +
  41 The `register` subcomand will also set the `maintenance.strategy` config
  42 value to `incremental`, if this value is not previously set. The
  43 `incremental` strategy uses the following schedule for each maintenance
  44 task:
  45 +
  46 --
  47 * `gc`: disabled.
  48 * `commit-graph`: hourly.
  49 * `prefetch`: hourly.
  50 * `loose-objects`: daily.
  51 * `incremental-repack`: daily.
  52 --
  53 +
  54 `git maintenance register` will also disable foreground maintenance by
  55 setting `maintenance.auto = false` in the current repository. This config
  56 setting will remain after a `git maintenance unregister` command.
  57
  58 run::
  59         Run one or more maintenance tasks. If one or more `--task` options
  60         are specified, then those tasks are run in that order. Otherwise,
  61         the tasks are determined by which `maintenance.<task>.enabled`
  62         config options are true. By default, only `maintenance.gc.enabled`
  63         is true.
  64
  65 start::
  66         Start running maintenance on the current repository. This performs
  67         the same config updates as the `register` subcommand, then updates
  68         the background scheduler to run `git maintenance run --scheduled`
  69         on an hourly basis.
  70
  71 stop::
  72         Halt the background maintenance schedule. The current repository
  73         is not removed from the list of maintained repositories, in case
  74         the background maintenance is restarted later.
  75
  76 unregister::
  77         Remove the current repository from background maintenance. This
  78         only removes the repository from the configured list. It does not
  79         stop the background maintenance processes from running.
  80
  81 TASKS
  82 -----
  83
  84 commit-graph::
  85         The `commit-graph` job updates the `commit-graph` files incrementally,
  86         then verifies that the written data is correct. The incremental
  87         write is safe to run alongside concurrent Git processes since it
  88         will not expire `.graph` files that were in the previous
  89         `commit-graph-chain` file. They will be deleted by a later run based
  90         on the expiration delay.
  91
  92 prefetch::
  93         The `prefetch` task updates the object directory with the latest
  94         objects from all registered remotes. For each remote, a `git fetch`
  95         command is run. The refmap is custom to avoid updating local or remote
  96         branches (those in `refs/heads` or `refs/remotes`). Instead, the
  97         remote refs are stored in `refs/prefetch/<remote>/`. Also, tags are
  98         not updated.
  99 +
 100 This is done to avoid disrupting the remote-tracking branches. The end users
 101 expect these refs to stay unmoved unless they initiate a fetch.  With prefetch
 102 task, however, the objects necessary to complete a later real fetch would
 103 already be obtained, so the real fetch would go faster.  In the ideal case,
 104 it will just become an update to bunch of remote-tracking branches without
 105 any object transfer.
 106
 107 gc::
 108         Clean up unnecessary files and optimize the local repository. "GC"
 109         stands for "garbage collection," but this task performs many
 110         smaller tasks. This task can be expensive for large repositories,
 111         as it repacks all Git objects into a single pack-file. It can also
 112         be disruptive in some situations, as it deletes stale data. See
 113         linkgit:git-gc[1] for more details on garbage collection in Git.
 114
 115 loose-objects::
 116         The `loose-objects` job cleans up loose objects and places them into
 117         pack-files. In order to prevent race conditions with concurrent Git
 118         commands, it follows a two-step process. First, it deletes any loose
 119         objects that already exist in a pack-file; concurrent Git processes
 120         will examine the pack-file for the object data instead of the loose
 121         object. Second, it creates a new pack-file (starting with "loose-")
 122         containing a batch of loose objects. The batch size is limited to 50
 123         thousand objects to prevent the job from taking too long on a
 124         repository with many loose objects. The `gc` task writes unreachable
 125         objects as loose objects to be cleaned up by a later step only if
 126         they are not re-added to a pack-file; for this reason it is not
 127         advisable to enable both the `loose-objects` and `gc` tasks at the
 128         same time.
 129
 130 incremental-repack::
 131         The `incremental-repack` job repacks the object directory
 132         using the `multi-pack-index` feature. In order to prevent race
 133         conditions with concurrent Git commands, it follows a two-step
 134         process. First, it calls `git multi-pack-index expire` to delete
 135         pack-files unreferenced by the `multi-pack-index` file. Second, it
 136         calls `git multi-pack-index repack` to select several small
 137         pack-files and repack them into a bigger one, and then update the
 138         `multi-pack-index` entries that refer to the small pack-files to
 139         refer to the new pack-file. This prepares those small pack-files
 140         for deletion upon the next run of `git multi-pack-index expire`.
 141         The selection of the small pack-files is such that the expected
 142         size of the big pack-file is at least the batch size; see the
 143         `--batch-size` option for the `repack` subcommand in
 144         linkgit:git-multi-pack-index[1]. The default batch-size is zero,
 145         which is a special case that attempts to repack all pack-files
 146         into a single pack-file.
 147
 148 OPTIONS
 149 -------
 150 --auto::
 151         When combined with the `run` subcommand, run maintenance tasks
 152         only if certain thresholds are met. For example, the `gc` task
 153         runs when the number of loose objects exceeds the number stored
 154         in the `gc.auto` config setting, or when the number of pack-files
 155         exceeds the `gc.autoPackLimit` config setting. Not compatible with
 156         the `--schedule` option.
 157
 158 --schedule::
 159         When combined with the `run` subcommand, run maintenance tasks
 160         only if certain time conditions are met, as specified by the
 161         `maintenance.<task>.schedule` config value for each `<task>`.
 162         This config value specifies a number of seconds since the last
 163         time that task ran, according to the `maintenance.<task>.lastRun`
 164         config value. The tasks that are tested are those provided by
 165         the `--task=<task>` option(s) or those with
 166         `maintenance.<task>.enabled` set to true.
 167
 168 --quiet::
 169         Do not report progress or other information over `stderr`.
 170
 171 --task=<task>::
 172         If this option is specified one or more times, then only run the
 173         specified tasks in the specified order. If no `--task=<task>`
 174         arguments are specified, then only the tasks with
 175         `maintenance.<task>.enabled` configured as `true` are considered.
 176         See the 'TASKS' section for the list of accepted `<task>` values.
 177
 178
 179 TROUBLESHOOTING
 180 ---------------
 181 The `git maintenance` command is designed to simplify the repository
 182 maintenance patterns while minimizing user wait time during Git commands.
 183 A variety of configuration options are available to allow customizing this
 184 process. The default maintenance options focus on operations that complete
 185 quickly, even on large repositories.
 186
 187 Users may find some cases where scheduled maintenance tasks do not run as
 188 frequently as intended. Each `git maintenance run` command takes a lock on
 189 the repository's object database, and this prevents other concurrent
 190 `git maintenance run` commands from running on the same repository. Without
 191 this safeguard, competing processes could leave the repository in an
 192 unpredictable state.
 193
 194 The background maintenance schedule runs `git maintenance run` processes
 195 on an hourly basis. Each run executes the "hourly" tasks. At midnight,
 196 that process also executes the "daily" tasks. At midnight on the first day
 197 of the week, that process also executes the "weekly" tasks. A single
 198 process iterates over each registered repository, performing the scheduled
 199 tasks for that frequency. Depending on the number of registered
 200 repositories and their sizes, this process may take longer than an hour.
 201 In this case, multiple `git maintenance run` commands may run on the same
 202 repository at the same time, colliding on the object database lock. This
 203 results in one of the two tasks not running.
 204
 205 If you find that some maintenance windows are taking longer than one hour
 206 to complete, then consider reducing the complexity of your maintenance
 207 tasks. For example, the `gc` task is much slower than the
 208 `incremental-repack` task. However, this comes at a cost of a slightly
 209 larger object database. Consider moving more expensive tasks to be run
 210 less frequently.
 211
 212 Expert users may consider scheduling their own maintenance tasks using a
 213 different schedule than is available through `git maintenance start` and
 214 Git configuration options. These users should be aware of the object
 215 database lock and how concurrent `git maintenance run` commands behave.
 216 Further, the `git gc` command should not be combined with
 217 `git maintenance run` commands. `git gc` modifies the object database
 218 but does not take the lock in the same way as `git maintenance run`. If
 219 possible, use `git maintenance run --task=gc` instead of `git gc`.
 220
 221
 222 GIT
 223 ---
 224 Part of the linkgit:git[1] suite