docs/AUTOMATIC_BOOT_ASSESSMENT.md

   1 ---
   2 title: Automatic Boot Assessment
   3 category: Booting
   4 ---
   5
   6 # Automatic Boot Assessment
   7
   8 systemd provides support for automatically reverting back to the previous
   9 version of the OS or kernel in case the system consistently fails to boot. This
  10 support is built into various of its components. When used together these
  11 components provide a complete solution on UEFI systems, built as add-on to the
  12 [Boot Loader
  13 Specification](https://systemd.io/BOOT_LOADER_SPECIFICATION). However, the
  14 different components may also be used independently, and in combination with
  15 other software, to implement similar schemes, for example with other boot
  16 loaders or for non-UEFI systems. Here's a brief overview of the complete set of
  17 components:
  18
  19 * The
  20   [`systemd-boot(7)`](https://www.freedesktop.org/software/systemd/man/systemd-boot.html)
  21   boot loader optionally maintains a per-boot-loader-entry counter that is
  22   decreased by one on each attempt to boot the entry, prioritizing entries that
  23   have non-zero counters over those which already reached a counter of zero
  24   when choosing the entry to boot.
  25
  26 * The
  27   [`systemd-bless-boot.service(8)`](https://www.freedesktop.org/software/systemd/man/systemd-bless-boot.service.html)
  28   service automatically marks a boot loader entry, for which boot counting as
  29   mentioned above is enabled, as "good" when a boot has been determined to be
  30   successful, thus turning off boot counting for it.
  31
  32 * The
  33   [`systemd-bless-boot-generator(8)`](https://www.freedesktop.org/software/systemd/man/systemd-bless-boot-generator.html)
  34   generator automatically pulls in `systemd-bless-boot.service` when use of
  35   `systemd-boot` with boot counting enabled is detected.
  36
  37 * The
  38   [`systemd-boot-check-no-failures.service(8)`](https://www.freedesktop.org/software/systemd/man/systemd-boot-check-no-failures.service.html)
  39   service is a simple health check tool that determines whether the boot
  40   completed successfully. When enabled it becomes an indirect dependency of
  41   `systemd-bless-boot.service` (by means of `boot-complete.target`, see
  42   below), ensuring that the boot will not be considered successful if there are
  43   any failed services.
  44
  45 * The `boot-complete.target` target unit (see
  46   [`systemd.special(7)`](https://www.freedesktop.org/software/systemd/man/systemd.special.html))
  47   serves as a generic extension point both for units that shall be considered
  48   necessary to consider a boot successful on one side (example:
  49   `systemd-boot-check-no-failures.service` as described above), and units that
  50   want to act only if the boot is successful on the other (example:
  51   `systemd-bless-boot.service` as described above).
  52
  53 * The
  54   [`kernel-install(8)`](https://www.freedesktop.org/software/systemd/man/kernel-install.html)
  55   script can optionally create boot loader entries that carry an initial boot
  56   counter (the initial counter is configurable in `/etc/kernel/tries`).
  57
  58 # Details
  59
  60 The boot counting data `systemd-boot` and `systemd-bless-boot.service`
  61 manage is stored in the name of the boot loader entries. If a boot loader entry
  62 file name contains `+` followed by one or two numbers (if two numbers, then
  63 those need to be separated by `-`) right before the `.conf` suffix, then boot
  64 counting is enabled for it. The first number is the "tries left" counter
  65 encoding how many attempts to boot this entry shall still be made. The second
  66 number is the "tries done" counter, encoding how many failed attempts to boot
  67 it have already been made. Each time a boot loader entry marked this way is
  68 booted the first counter is decreased by one, and the second one increased by
  69 one. (If the second counter is missing, then it is assumed to be equivalent to
  70 zero.) If the "tries left" counter is above zero the entry is still considered
  71 for booting (the entry's state is considered to be "indeterminate"), as soon as
  72 it reached zero the entry is not tried anymore (entry state "bad"). If the boot
  73 attempt completed successfully the entry's counters are removed from the name
  74 (entry state "good"), thus turning off boot counting for the future.
  75
  76 ## Walkthrough
  77
  78 Here's an example walkthrough of how this all fits together.
  79
  80 1. The user runs `echo 3 > /etc/kernel/tries` to enable boot counting.
  81
  82 2. A new kernel is installed. `kernel-install` is used to generate a new boot
  83    loader entry file for it. Let's say the version string for the new kernel is
  84    `4.14.11-300.fc27.x86_64`, a new boot loader entry
  85    `/boot/loader/entries/4.14.11-300.fc27.x86_64+3.conf` is hence created.
  86
  87 3. The system is booted for the first time after the new kernel is
  88    installed. The boot loader now sees the `+3` counter in the entry file
  89    name. It hence renames the file to `4.14.11-300.fc27.x86_64+2-1.conf`
  90    indicating that at this point one attempt has started and thus only one less
  91    is left. After the rename completed the entry is booted as usual.
  92
  93 4. Let's say this attempt to boot fails. On the following boot the boot loader
  94    will hence see the `+2-1` tag in the name, and hence rename the entry file to
  95    `4.14.11-300.fc27.x86_64+1-2.conf`, and boot it.
  96
  97 5. Let's say the boot fails again. On the subsequent boot the loader hence will
  98    see the `+1-2` tag, and rename the file to
  99    `4.14.11-300.fc27.x86_64+0-3.conf` and boot it.
 100
 101 6. If this boot also fails, on the next boot the boot loader will see the
 102    tag `+0-3`, i.e. the counter reached zero. At this point the entry will be
 103    considered "bad", and ordered to the end of the list of entries. The next
 104    newest boot entry is now tried, i.e. the system automatically reverted back
 105    to an earlier version.
 106
 107 The above describes the walkthrough when the selected boot entry continuously
 108 fails. Let's have a look at an alternative ending to this walkthrough. In this
 109 scenario the first 4 steps are the same as above:
 110
 111 1. *as above*
 112
 113 2. *as above*
 114
 115 3. *as above*
 116
 117 4. *as above*
 118
 119 5. Let's say the second boot succeeds. The kernel initializes properly, systemd
 120    is started and invokes all generators.
 121
 122 6. One of the generators started is `systemd-bless-boot-generator` which
 123    detects that boot counting is used. It hence pulls
 124    `systemd-bless-boot.service` into the initial transaction.
 125
 126 7. `systemd-bless-boot.service` is ordered after and `Requires=` the generic
 127    `boot-complete.target` unit. This unit is hence also pulled into the initial
 128    transaction.
 129
 130 8. The `boot-complete.target` unit is ordered after and pulls in various units
 131    that are required to succeed for the boot process to be considered
 132    successful. One such unit is `systemd-boot-check-no-failures.service`.
 133
 134 9. `systemd-boot-check-no-failures.service` is run after all its own
 135    dependencies completed, and assesses that the boot completed
 136    successfully. It hence exits cleanly.
 137
 138 10. This allows `boot-complete.target` to be reached. This signifies to the
 139     system that this boot attempt shall be considered successful.
 140
 141 11. Which in turn permits `systemd-bless-boot.service` to run. It now
 142     determines which boot loader entry file was used to boot the system, and
 143     renames it dropping the counter tag. Thus
 144     `4.14.11-300.fc27.x86_64+1-2.conf` is renamed to
 145     `4.14.11-300.fc27.x86_64.conf`. From this moment boot counting is turned
 146     off.
 147
 148 12. On the following boot (and all subsequent boots after that) the entry is
 149     now seen with boot counting turned off, no further renaming takes place.
 150
 151 # How to adapt this scheme to other setups
 152
 153 Of the stack described above many components may be replaced or augmented. Here
 154 are a couple of recommendations.
 155
 156 1. To support alternative boot loaders in place of `systemd-boot` two scenarios
 157    are recommended:
 158
 159     a. Boot loaders already implementing the Boot Loader Specification can simply
 160        implement an equivalent file rename based logic, and thus integrate fully
 161        with the rest of the stack.
 162
 163     b. Boot loaders that want to implement boot counting and store the counters
 164        elsewhere can provide their own replacements for
 165        `systemd-bless-boot.service` and `systemd-bless-boot-generator`, but should
 166        continue to use `boot-complete.target` and thus support any services
 167        ordered before that.
 168
 169 2. To support additional components that shall succeed before the boot is
 170    considered successful, simply place them in units (if they aren't already)
 171    and order them before the generic `boot-complete.target` target unit,
 172    combined with `Requires=` dependencies from the target, so that the target
 173    cannot be reached when any of the units fail. You may add any number of
 174    units like this, and only if they all succeed the boot entry is marked as
 175    good. Note that the target unit shall pull in these boot checking units, not
 176    the other way around.
 177
 178 3. To support additional components that shall only run on boot success, simply
 179    wrap them in a unit and order them after `boot-complete.target`, pulling it
 180    in.
 181
 182 # FAQ
 183
 184 1. *Why do you use file renames to store the counter? Why not a regular file?*
 185    — Mainly two reasons: it's relatively likely that renames can be implemented
 186    atomically even in simpler file systems, while writing to file contents has
 187    a much bigger chance to be result in incomplete or corrupt data, as renaming
 188    generally avoids allocating or releasing data blocks. Moreover it has the
 189    benefit that the boot count metadata is directly attached to the boot loader
 190    entry file, and thus the lifecycle of the metadata and the entry itself are
 191    bound together. This means no additional clean-up needs to take place to
 192    drop the boot loader counting information for an entry when it is removed.
 193
 194 2. *Why not use EFI variables for storing the boot counter?* — The memory chips
 195    used to back the persistent EFI variables are generally not of the highest
 196    quality, hence shouldn't be written to more than necessary. This means we
 197    can't really use it for changes made regularly during boot, but can use it
 198    only for seldom made configuration changes.
 199
 200 3. *I have a service which — when it fails — should immediately cause a
 201    reboot. How does that fit in with the above?* — Well, that's orthogonal to
 202    the above, please use `FailureAction=` in the unit file for this.
 203
 204 4. *Under some condition I want to mark the current boot loader entry as bad
 205    right-away, so that it never is tried again, how do I do that?* — You may
 206    invoke `/usr/lib/systemd/systemd-bless-boot bad` at any time to mark the
 207    current boot loader entry as "bad" right-away so that it isn't tried again
 208    on later boots.