You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
fog-agent unable to reboot a Linux computer due to active block inhibitor #30
Describe the bug
When the fog-agent is installed on a Linux computer that has an active systemd inhibitor (systemd-inhibit --list), it will be unable to be rebooted.
To Reproduce
Steps to reproduce the behavior:
Install the fog-agent onto Fedora Workstation
Run a deploy task from FOG
Check the fog-agent.service logs on your client after it receives the request
See it output reboot: Failed to schedule shutdown: Operation denied due to active block inhibitor
Expected behavior
The fog-agent should force a reboot regardless
Screenshots
N/A
Software (please complete the following information):
FOG version: 1.6 RC 11
OS: Rocky Linux 10.2
Additional context
Unsure if Windows clients have a similar issue
Also unsure which repo to put this in, as apologies if I should have used the fog-agent one
Thanks for the report. I moved it here from the fogproject repo, since the problem is in the agent.
Cause: logind enforces systemd block inhibitor locks. The agent reboots through logind (shutdown +N or systemctl reboot), so logind refused it. The agent then retried on every poll and never rebooted. Root also cannot skip the lock with systemctl -i, because logind then asks polkit for an interactive answer.
Fix: v0.1.16 (#31). When logind refuses, the agent now starts reboot.target directly through systemd. This is the same orderly reboot logind runs when it allows one. If a reboot has a warning delay, the agent creates fog-agent-reboot.timer for the countdown and sends the warning with wall. To cancel it, run systemctl stop fog-agent-reboot.timer. Windows already worked this way with shutdown /f.
To get it: an agent on the "latest" update ring updates itself to v0.1.16. You can also install it from the v0.1.16 release.
Can you tell us which program held the lock? Run systemd-inhibit --list while the reboot is blocked and paste the output. A package manager in the middle of a transaction would be cut off by the new behavior, and we want to know if that is the case you hit.
ModemManager 0 root 1138 ModemManager sleep ModemManager needs to reset devices delay
NetworkManager 0 root 1233 NetworkManager sleep NetworkManager needs to turn off networks delay
UPower 0 root 1050 upowerd sleep Pause device polling delay
GNOME Shell 1000 user 3340 gnome-shell sleep GNOME needs to lock the screen delay
GNOME Shell 1000 user 3340 gnome-shell sleep GNOME needs to save screen time data delay
user 1000 user 3481 gsd-power handle-lid-switch External monitor attached or configuration changed recently block
user 1000 user 3480 gsd-media-keys handle-power-key:handle-suspend-key:handle-hibernate-key GNOME handling keypresses block
user 1000 user 3316 gnome-session-s shutdown user session inhibited block
user 1000 user 3480 gsd-media-keys sleep GNOME handling keypresses delay
user 1000 user 3481 gsd-power sleep GNOME needs to lock the screen delay
However, we are running into a new issue:
Our entirely stock Fedora 44 Workstation is not updating the client. All of our Windows 11 images were able to. There was no logs on the machine itself from journalctl or systemctl and history items > agent activity showed nothing about updating. I'm unsure if this is a known issue, so just saying here for now.
Also if there is a more preferred communication channel for this, please let me know
Activity
mastacontrola commentedon Oct 14, 2026
Thanks for the report. I moved it here from the fogproject repo, since the problem is in the agent.
Cause: logind enforces systemd block inhibitor locks. The agent reboots through logind (
shutdown +Norsystemctl reboot), so logind refused it. The agent then retried on every poll and never rebooted. Root also cannot skip the lock withsystemctl -i, because logind then asks polkit for an interactive answer.Fix: v0.1.16 (#31). When logind refuses, the agent now starts
reboot.targetdirectly through systemd. This is the same orderly reboot logind runs when it allows one. If a reboot has a warning delay, the agent createsfog-agent-reboot.timerfor the countdown and sends the warning withwall. To cancel it, runsystemctl stop fog-agent-reboot.timer. Windows already worked this way withshutdown /f.To get it: an agent on the "latest" update ring updates itself to v0.1.16. You can also install it from the v0.1.16 release.
Can you tell us which program held the lock? Run
systemd-inhibit --listwhile the reboot is blocked and paste the output. A package manager in the middle of a transaction would be cut off by the new behavior, and we want to know if that is the case you hit.Sempyos commentedon Oct 14, 2026
It was just the gnome session
Sempyos commentedon Oct 14, 2026
Tested the new version, works perfectly
However, we are running into a new issue:
Our entirely stock Fedora 44 Workstation is not updating the client. All of our Windows 11 images were able to. There was no logs on the machine itself from
journalctlorsystemctland history items > agent activity showed nothing about updating. I'm unsure if this is a known issue, so just saying here for now.Also if there is a more preferred communication channel for this, please let me know
Sempyos commentedon Oct 14, 2026
Nevermind that update issue, seems something caused the agent to just lock up, rebooting twice fixed it