systemd provides support for automatically reverting back to the previous version of the OS or kernel in case the system consistently fails to boot. This support is built into various of its components. When used together these components provide a complete solution on UEFI systems, built as add-on to the Boot Loader Specification. However, the different components may also be used independently, and in combination with other software, to implement similar schemes, for example with other boot loaders or for non-UEFI systems. Here's a brief overview of the complete set of components:
The systemd-boot(7) boot loader optionally maintains a per-boot-loader-entry counter that is decreased by one on each attempt to boot the entry, prioritizing entries that have non-zero counters over those which already reached a counter of zero when choosing the entry to boot.
The systemd-bless-boot.service(8) service automatically marks a boot loader entry, for which boot counting as mentioned above is enabled, as “good” when a boot has been determined to be successful, thus turning off boot counting for it.
The systemd-bless-boot-generator(8) generator automatically pulls in systemd-bless-boot.service when use of systemd-boot with boot counting enabled is detected.
The systemd-boot-check-no-failures.service(8) service is a simple health check tool that determines whether the boot completed successfully. When enabled it becomes an indirect dependency of systemd-bless-boot.service (by means of boot-complete.target, see below), ensuring that the boot will not be considered successful if there are any failed services.
The boot-complete.target target unit (see systemd.special(7)) serves as a generic extension point both for units that shall be considered necessary to consider a boot successful on one side (example: systemd-boot-check-no-failures.service as described above), and units that want to act only if the boot is successful on the other (example: systemd-bless-boot.service as described above).
The kernel-install(8) script can optionally create boot loader entries that carry an initial boot counter (the initial counter is configurable in /etc/kernel/tries).
The boot counting data systemd-boot and systemd-bless-boot.service manage is stored in the name of the boot loader entries. If a boot loader entry file name contains + followed by one or two numbers (if two numbers, then those need to be separated by -) right before the .conf suffix, then boot counting is enabled for it. The first number is the “tries left” counter encoding how many attempts to boot this entry shall still be made. The second number is the “tries done” counter, encoding how many failed attempts to boot it have already been made. Each time a boot loader entry marked this way is booted the first counter is decreased by one, and the second one increased by one. (If the second counter is missing, then it is assumed to be equivalent to zero.) If the “tries left” counter is above zero the entry is still considered for booting (the entry‘s state is considered to be “indeterminate”), as soon as it reached zero the entry is not tried anymore (entry state “bad”). If the boot attempt completed successfully the entry’s counters are removed from the name (entry state “good”), thus turning off boot counting for the future.
Here's an example walkthrough of how this all fits together.
The user runs echo 3 > /etc/kernel/tries to enable boot counting.
A new kernel is installed. kernel-install is used to generate a new boot loader entry file for it. Let's say the version string for the new kernel is 4.14.11-300.fc27.x86_64, a new boot loader entry /boot/loader/entries/4.14.11-300.fc27.x86_64+3.conf is hence created.
The system is booted for the first time after the new kernel is installed. The boot loader now sees the +3 counter in the entry file name. It hence renames the file to 4.14.11-300.fc27.x86_64+2-1.conf indicating that at this point one attempt has started and thus only one less is left. After the rename completed the entry is booted as usual.
Let's say this attempt to boot fails. On the following boot the boot loader will hence see the +2-1 tag in the name, and hence rename the entry file to 4.14.11-300.fc27.x86_64+1-2.conf, and boot it.
Let's say the boot fails again. On the subsequent boot the loader hence will see the +1-2 tag, and rename the file to 4.14.11-300.fc27.x86_64+0-3.conf and boot it.
If this boot also fails, on the next boot the boot loader will see the tag +0-3, i.e. the counter reached zero. At this point the entry will be considered “bad”, and ordered to the beginning of the list of entries. The next newest boot entry is now tried, i.e. the system automatically reverted back to an earlier version.
The above describes the walkthrough when the selected boot entry continuously fails. Let's have a look at an alternative ending to this walkthrough. In this scenario the first 4 steps are the same as above:
as above
as above
as above
as above
Let's say the second boot succeeds. The kernel initializes properly, systemd is started and invokes all generators.
One of the generators started is systemd-bless-boot-generator which detects that boot counting is used. It hence pulls systemd-bless-boot.service into the initial transaction.
systemd-bless-boot.service is ordered after and Requires= the generic boot-complete.target unit. This unit is hence also pulled into the initial transaction.
The boot-complete.target unit is ordered after and pulls in various units that are required to succeed for the boot process to be considered successful. One such unit is systemd-boot-check-no-failures.service.
systemd-boot-check-no-failures.service is run after all its own dependencies completed, and assesses that the boot completed successfully. It hence exits cleanly.
This allows boot-complete.target to be reached. This signifies to the system that this boot attempt shall be considered successful.
Which in turn permits systemd-bless-boot.service to run. It now determines which boot loader entry file was used to boot the system, and renames it dropping the counter tag. Thus 4.14.11-300.fc27.x86_64+1-2.conf is renamed to 4.14.11-300.fc27.x86_64.conf. From this moment boot counting is turned off.
On the following boot (and all subsequent boots after that) the entry is now seen with boot counting turned off, no further renaming takes place.
Of the stack described above many components may be replaced or augmented. Here are a couple of recommendations.
To support alternative boot loaders in place of systemd-boot two scenarios are recommended:
a. Boot loaders already implementing the Boot Loader Specification can simply implement an equivalent file rename based logic, and thus integrate fully with the rest of the stack.
b. Boot loaders that want to implement boot counting and store the counters elsewhere can provide their own replacements for systemd-bless-boot.service and systemd-bless-boot-generator, but should continue to use boot-complete.target and thus support any services ordered before that.
To support additional components that shall succeed before the boot is considered successful, simply place them in units (if they aren't already) and order them before the generic boot-complete.target target unit, combined with Requires= dependencies from the target, so that the target cannot be reached when any of the units fail. You may add any number of units like this, and only if they all succeed the boot entry is marked as good. Note that the target unit shall pull in these boot checking units, not the other way around.
To support additional components that shall only run on boot success, simply wrap them in a unit and order them after boot-complete.target, pulling it in.
Why do you use file renames to store the counter? Why not a regular file? — Mainly two reasons: it's relatively likely that renames can be implemented atomically even in simpler file systems, while writing to file contents has a much bigger chance to be result in incomplete or corrupt data, as renaming generally avoids allocating or releasing data blocks. Moreover it has the benefit that the boot count metadata is directly attached to the boot loader entry file, and thus the lifecycle of the metadata and the entry itself are bound together. This means no additional clean-up needs to take place to drop the boot loader counting information for an entry when it is removed.
Why not use EFI variables for storing the boot counter? — The memory chips used to back the persistent EFI variables are generally not of the highest quality, hence shouldn‘t be written to more than necessary. This means we can’t really use it for changes made regularly during boot, but can use it only for seldom made configuration changes.
I have a service which — when it fails — should immediately cause a reboot. How does that fit in with the above? — Well, that's orthogonal to the above, please use FailureAction= in the unit file for this.
Under some condition I want to mark the current boot loader entry as bad right-away, so that it never is tried again, how do I do that? — You may invoke /usr/lib/systemd/systemd-bless-boot bad at any time to mark the current boot loader entry as “bad” right-away so that it isn't tried again on later boots.