mirror of
https://github.com/webmin/webmin.git
synced 2026-06-10 14:50:33 +01:00
a83db9f57d
This PR expands the "Bootup and Shutdown" module’s systemd support with creation and management for services, timers, sockets, paths and targets. It also adds user-scoped systemd units, linger controls, status and log actions, tabbed unit listings by type, and contextual help for the new options. For user-scoped units, the implementation includes several safety guards because unit files live under user-controlled home directories: - User accounts are validated with system account data before any user-unit operation is attempted. - User unit names are restricted to known systemd unit suffixes and safe filename characters. - User unit files are limited to direct children of `~/.config/systemd/user`. - Symlinked `.config`, `.config/systemd`, and `.config/systemd/user` paths are rejected. - User unit reads, writes, directory creation, and deletes are performed after dropping privileges to the target Unix user. - File operations re-check paths close to the actual read/write/delete operation to reduce symlink race exposure. - User unit create failures roll back half-created files when daemon reload fails. - User-provided unit names, owners, paths, command output, and logs are HTML-escaped before display. - systemctl, journalctl, and loginctl command arguments are shell-quoted before execution. - User services omit `User=` and `Group=` directives because they already run under the selected user’s systemd manager. Together, all these changes will allow Webmin admin to manage both system and user systemd units while keeping user-controlled home-directory paths from becoming root-level file read/write/delete exploits. Implemented in response to these two issue requests https://github.com/webmin/webmin/issues/2733 and https://github.com/webmin/webmin/issues/2734
10 lines
554 B
HTML
10 lines
554 B
HTML
<header>Restart policy</header>
|
|
<p>Controls when systemd restarts the service after it exits. This writes
|
|
<tt>Restart=</tt>.</p>
|
|
<p><tt>no</tt> disables automatic restarts. <tt>on-failure</tt> restarts after
|
|
non-zero exits, signals, timeouts, and watchdog failures. <tt>always</tt>
|
|
restarts after almost any exit except an explicit stop by systemd.</p>
|
|
<p><tt>on-success</tt>, <tt>on-abnormal</tt>, <tt>on-abort</tt>, and
|
|
<tt>on-watchdog</tt> are narrower policies. For ordinary server processes,
|
|
<tt>on-failure</tt> is usually the practical choice.</p>
|