systemd/man/systemd.offline-updates.xml
2023-12-25 15:51:47 +01:00

169 lines
8.8 KiB
XML

<?xml version='1.0'?> <!--*-nxml-*-->
<!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook XML V4.5//EN"
"http://www.oasis-open.org/docbook/xml/4.5/docbookx.dtd">
<!-- SPDX-License-Identifier: LGPL-2.1-or-later -->
<refentry id="systemd.offline-updates">
<refentryinfo>
<title>systemd.offline-updates</title>
<productname>systemd</productname>
</refentryinfo>
<refmeta>
<refentrytitle>systemd.offline-updates</refentrytitle>
<manvolnum>7</manvolnum>
</refmeta>
<refnamediv>
<refname>systemd.offline-updates</refname>
<refpurpose>Implementation of offline updates in systemd</refpurpose>
</refnamediv>
<refsect1>
<title>Implementing Offline System Updates</title>
<para>This man page describes how to implement "offline" system updates with systemd. By "offline"
OS updates we mean package installations and updates that are run with the system booted into a
special system update mode, in order to avoid problems related to conflicts of libraries and
services that are currently running with those on disk. This document is inspired by this
<ulink url="https://wiki.gnome.org/Design/OS/SoftwareUpdates">GNOME design whiteboard</ulink>.
</para>
<para>The logic:</para>
<orderedlist>
<listitem>
<para>The package manager prepares system updates by downloading all (.rpm or .deb or
whatever) packages to update off-line in a special directory
<filename index="false">/var/lib/system-update</filename> (or
another directory of the package/upgrade manager's choice).</para>
</listitem>
<listitem>
<para>When the user OK'ed the update, the symlink <filename>/system-update</filename> or
<filename>/etc/system-update</filename> is created that points to
<filename index="false">/var/lib/system-update</filename> (or wherever the directory with
the upgrade files is located) and the system is rebooted. This symlink is in the root
directory, since we need to check for it very early at boot, at a time where
<filename>/var/</filename> is not available yet.</para>
</listitem>
<listitem>
<para>Very early in the new boot
<citerefentry><refentrytitle>systemd-system-update-generator</refentrytitle><manvolnum>8</manvolnum></citerefentry>
checks whether <filename>/system-update</filename> or
<filename>/etc/system-update</filename> exists. If so, it (temporarily and for this boot
only) redirects (i.e. symlinks) <filename>default.target</filename> to
<filename>system-update.target</filename>, a special target that pulls in the base system
(i.e. <filename>sysinit.target</filename>, so that all file systems are mounted but little
else) and the system update units.</para>
</listitem>
<listitem>
<para>The system now continues to boot into <filename>default.target</filename>, and
thus into <filename>system-update.target</filename>. This target pulls in all system
update units. Only one service should perform an update (see the next point), and all
the other ones should exit cleanly with a "success" return code and without doing
anything. Update services should be ordered after <filename>sysinit.target</filename>
so that the update starts after all file systems have been mounted.</para>
</listitem>
<listitem>
<para>As the first step, an update service should check if the
<filename>/system-update</filename> or <filename>/etc/system-update</filename> symlink
points to the location used by that update service. In case it does not exist or points to a
different location, the service must exit without error. It is possible for multiple update
services to be installed, and for multiple update services to be launched in parallel, and
only the one that corresponds to the tool that <emphasis>created</emphasis> the symlink
before reboot should perform any actions. It is unsafe to run multiple updates in
parallel.</para>
</listitem>
<listitem>
<para>The update service should now do its job. If applicable and possible, it should
create a file system snapshot, then install all packages. After completion (regardless
whether the update succeeded or failed) the machine must be rebooted, for example by
calling <command>systemctl reboot</command>. In addition, on failure the script should
revert to the old file system snapshot (without the symlink).</para>
</listitem>
<listitem>
<para>The update scripts should exit only after the update is finished. It is expected
that the service which performs the update will cause the machine to reboot after it
is done. If the <filename>system-update.target</filename> is successfully reached, i.e.
all update services have run, and the <filename>/system-update</filename> or
<filename>/etc/system-update</filename> symlink still exists, it will be removed and
the machine rebooted as a safety measure.</para>
</listitem>
<listitem>
<para>After a reboot, now that the <filename>/system-update</filename> and
<filename>/etc/system-update</filename> symlink is gone, the generator won't redirect
<filename>default.target</filename> anymore and the system now boots into the default
target again.</para>
</listitem>
</orderedlist>
</refsect1>
<refsect1>
<title>Recommendations</title>
<orderedlist>
<listitem>
<para>To make things a bit more robust we recommend hooking the update script into
<filename>system-update.target</filename> via a <filename index="false">.wants/</filename>
symlink in the distribution package, rather than depending on <command>systemctl
enable</command> in the postinst scriptlets of your package. More specifically, for your
update script create a .service file, without [Install] section, and then add a symlink like
<filename index="false">/usr/lib/systemd/system/system-update.target.wants/foobar.service</filename>
<filename index="false">../foobar.service</filename> to your package.</para>
</listitem>
<listitem>
<para>Make sure to remove the <filename>/system-update</filename> and
<filename>/etc/system-update</filename> symlinks as early as possible in the update
script to avoid reboot loops in case the update fails.</para>
</listitem>
<listitem>
<para>Use <varname>FailureAction=reboot</varname> in the service file for your update script
to ensure that a reboot is automatically triggered if the update fails.
<varname>FailureAction=</varname> makes sure that the specified unit is activated if your
script exits uncleanly (by non-zero error code, or signal/coredump). If your script succeeds
you should trigger the reboot in your own code, for example by invoking logind's
<command>Reboot()</command> call or calling <command>systemctl reboot</command>. See
<citerefentry><refentrytitle>org.freedesktop.login1</refentrytitle><manvolnum>5</manvolnum></citerefentry>
for details about the logind D-Bus API.</para>
</listitem>
<listitem>
<para>The update service should declare <varname>DefaultDependencies=no</varname>,
<varname>Requires=sysinit.target</varname>, <varname>After=sysinit.target</varname>,
<varname>After=system-update-pre.target</varname>, <varname>Before=system-update.target</varname>
and explicitly pull in any other services it requires.</para>
</listitem>
<listitem>
<para>It may be desirable to always run an auxiliary unit when booting
into offline-updates mode, which itself does not install updates. To
do this create a .service file with
<varname>Wants=system-update-pre.target</varname> and
<varname>Before=system-update-pre.target</varname> and add a symlink
to that file under
<filename index="false">/usr/lib/systemd/system-update.target.wants</filename>
.</para>
</listitem>
</orderedlist>
</refsect1>
<refsect1>
<title>See Also</title>
<para><simplelist type="inline">
<member><citerefentry><refentrytitle>systemd</refentrytitle><manvolnum>1</manvolnum></citerefentry></member>
<member><citerefentry><refentrytitle>systemd.generator</refentrytitle><manvolnum>7</manvolnum></citerefentry></member>
<member><citerefentry><refentrytitle>systemd-system-update-generator</refentrytitle><manvolnum>8</manvolnum></citerefentry></member>
<member><citerefentry project='mankier'><refentrytitle>dnf.plugin.system-upgrade</refentrytitle><manvolnum>8</manvolnum></citerefentry></member>
</simplelist></para>
</refsect1>
</refentry>