Issues with Hibernation and Power State Change

Zeon · 2025-10-02 11:24:54

Hello community,

I have occasional problems with my hibernation. More rarely I noticed issues with shutdown as well.

I think my hibernation is set up correctly, because it works part of the time. Just sometimes it does not. In that case my system does not power off. I remember at least one situation when a regular shutdown didn't power off the system as well.

Observed Process

When choosing hibernate, my system shuts down UI (KDE Plasma), shows some lines on the terminal session, then my monitor loses signal. It takes some seconds. Then my sleeping HDDs spin up (I can hear that) (see below for background). A few seconds later the system powers off. That's when hibernate works. I can then cut the power and boot the system later and get my hibernated session resumed.

Sometimes, after the HDD spinup, the system stays in some half dead state. The power LED stays on. Fans stay on. HDDs stay on. Power button does nothing, if pressed shortly. Keyboard or mouse input does nothing. All black. I need to long-press the power button to force shutdown or cut the power. On boot, the session is lost.

I tried searching journalctl and dmesg for any info, but couldn't find any so far.

I appreciate any ideas on how to troubleshoot or fix this, because I need hibernation regularly. I chose hibernation, because I want to be able to resume my session later and want to be able to cut the power (flip the switch on my power strip) to my computer during sleep. As far as I know, all other standby/sleep methods require the power to stay available.

----

My sleeping HDDs

I have HDDs installed. I hate their idle noise. So I put them to sleep most of the time using hdparm -Y. KDE Plasma (I think or is it Dolphin?) wakes them up on login and they spin up shortly before power off. I just explain that here, because their noise helps me tracking the hibernation process (as described above) or rather describe, at which point things go wrong. I do not suspect them of making my hibernation fail.

Additional Information

uname -r: 6.16.8-arch3-1
I have a swap partition set up, slightly bigger than my RAM.
I'm using GRUB with a resume=UUID=... in /etc/default/grub.
resume hook is enabled in initramfs.

Last edited by Zeon (2025-10-02 11:32:41)

LuxFerre · 2025-10-02 16:56:28

Start with the end of journalctl -b -1 (or find the specific session where the crash happened, where -1 is the previous, -2 the one before that etc).
Enable sysrq (see arch wiki) and try it the next time it happens. Also try changing to a different TTY (ctrl alt F4). Cutting the power should be really last resort.

Disabling watchdogs might help as well.

Zeon · 2025-10-05 13:50:42

That's what I could find in journalctl after the last failed hibernation:

Oct 05 14:23:24 <host> systemd-logind[901]: The system will hibernate now!
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7225] manager: sleep: sleep requested (sleeping: no  enabled: yes)
Oct 05 14:23:24 <host> discord[1876]: 14:23:24.724 › [GatewaySocket] Setting connection state to WILL_RECONNECT
Oct 05 14:23:24 <host> discord[1876]: 14:23:24.724 › [GatewaySocket] [WS CLOSED] (false, 1006, ) retrying in 1.00 seconds.
Oct 05 14:23:24 <host> discord[1876]: 14:23:24.724 › [Spotify] WS Disconnected. Next retry in 762ms
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7227] device (wlp10s0): state change: unavailable -> unmanaged (reason 'unmanaged-sleeping', managed-type: 'full')
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7231] device (wlp10s0): set-hw-addr: reset MAC address to BC:38:98:D8:83:DA (unmanage)
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7235] manager: NetworkManager state is now ASLEEP
Oct 05 14:23:24 <host> dbus-broker[897]: A security policy denied :1.27 to send method call /org/freedesktop/login1/seat/seat0:org.freedesktop.login1.Seat.Inhibit to org.freedesktop.login1.
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7236] device (enp9s0): state change: activated -> deactivating (reason 'sleeping', managed-type: 'full')
Oct 05 14:23:24 <host> kwin_wayland[1216]: kwin_core: Failed to delay sleep: Sender is not authorized to send message
Oct 05 14:23:24 <host> dbus-broker[897]: A security policy denied :1.27 to send method call /org/freedesktop/login1/seat/seat0:org.freedesktop.login1.Seat.Inhibit to org.freedesktop.login1.
Oct 05 14:23:24 <host> kwin_wayland[1216]: kwin_core: Failed to delay sleep: Sender is not authorized to send message
Oct 05 14:23:24 <host> systemd[1]: Starting Network Manager Script Dispatcher Service...
Oct 05 14:23:24 <host> systemd[1]: Started Network Manager Script Dispatcher Service.
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7538] device (enp9s0): state change: deactivating -> disconnected (reason 'sleeping', managed-type: 'full')
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7541] dhcp4 (enp9s0): canceled DHCP transaction
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7541] dhcp4 (enp9s0): activation: beginning transaction (timeout in 45 seconds)
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7541] dhcp4 (enp9s0): state changed no lease
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7542] dhcp6 (enp9s0): canceled DHCP transaction
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7543] dhcp6 (enp9s0): activation: beginning transaction (timeout in 45 seconds)
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.7543] dhcp6 (enp9s0): state changed no lease
Oct 05 14:23:24 <host> discord[1876]: 14:23:24.755 › [GatewaySocket] Expedited heartbeat requested, but, connection state is WILL_RECONNECT and reconnectImmediately was not requested reason: network detected offline.
Oct 05 14:23:24 <host> NetworkManager[898]: <info>  [1759667004.8210] device (enp9s0): state change: disconnected -> unmanaged (reason 'unmanaged-sleeping', managed-type: 'full')
Oct 05 14:23:24 <host> kernel: r8169 0000:09:00.0 enp9s0: Link is Down
Oct 05 14:23:25 <host> systemd[1]: Reached target Sleep.
Oct 05 14:23:25 <host> systemd[1]: Starting System Hibernate...
Oct 05 14:23:25 <host> systemd[1]: user@1000.service: Unit now frozen-by-parent.
Oct 05 14:23:25 <host> systemd[1]: session-2.scope: Unit now frozen-by-parent.
Oct 05 14:23:25 <host> systemd[1]: user-1000.slice: Unit now frozen-by-parent.
Oct 05 14:23:25 <host> systemd[1]: user.slice: Unit now frozen.
Oct 05 14:23:25 <host> systemd-sleep[11330]: Successfully froze unit 'user.slice'.
Oct 05 14:23:25 <host> systemd-sleep[11330]: Performing sleep operation 'hibernate'...
Oct 05 14:23:25 <host> kernel: PM: hibernation: hibernation entry

I tried to switch TTY, but got no monitor signal anymore. I also connected the monitor to my mainboard to see if the signal maybe goes through there instead of my dedicated GPU, but no success either.

On powering on later, the system did not boot at all. I didn't get to GRUB. No monitor signal. Had to force power off again and power on. Then it booted (into a fresh session). I find that very weird, because if it were a faulty resume after hibernation, it would've happened after GRUB. I hope the hardware is not broken. Maybe a UEFI setting?

MB: Gigabyte B850 EAGLE
CPU: AMD Ryzen 9 9950X3D

I also got a dual boot setup, but Windows 11 is on its own SSD and also again, it should not fail before GRUB because of that.

Will try sysrq next time.

seth · 2025-10-05 21:43:11

I also got a dual boot setup, but Windows 11 is on its own SSD and also again, it should not fail before GRUB because of that.

3rd link below. Mandatory.
Disable it (it's NOT the BIOS setting!) and reboot windows and linux twice for voodo reasons.
(No, the own SSD doesn't matter for this condition)

If that doesn't fix it at all and while it kinda defeats the purpose of hibernation, can you try to reproduce this by logging out of plasma and hibernating from (I guess) SDDM?

Arch Linux

#1 2025-10-02 11:24:54