I’ve had two server oses here: alma linux and debian(currently). On both of them, they will hang when I shut them down from cockpit, and they hang at the end of the shutdown.
Also, it takes an hour to a day to have this issue start. if it’s restarted two times in a row quickly, it works perfectly fine for some reason.
What I’ve tried:
- setting “acpi=off” and “acpi=force” kernel parameters in grub
- removing my nvidia gpu(i was using nouveau drivers)
- changing distros
nothing worked. here are some things that both distros had in common with eachother:
- systemd
- cockpit
- libvirt & qemu
- docker
does anyone have advice? nothing i’ve seen online has worked. thank you for suggestions
reboot: machine restart
This makes me think it’s a motherboard issue.
The system is done with its shutdown process and issued the reboot command, but the motherboard didn’t restart.
There could be some electronics components which get wedged over time. My sound card will occasionally not boot unless it has been completely powered off for 30 seconds or so.
Hardware? Do they shut down properly if you do it from the console or ssh?
I will try to restart it using the
rebootcommand.The computer consists of:
- i3-10100
- 16gb ddr4 ram
- MSI h410m pro
Interesting. BIOS update? Maybe check through all the settings, or do a factory reset on the BIOS? I have a similar board (H510 something) running proxmox and it works fine.
I would like to note that this may have been caused by a bios update, as it started sometime after it. i’ll try another update now.
edit: already on the latest bios version.
@potentiallynotfelix As a diagnostic, I would suggest trying shutting them down by ssh in and then using systemctl to shut them down, if that works then you know the issue is with cockpit. If it hangs even when systemd is asked to halt then I would consider reverting to the previous bios and see if the problem persists.
Ok. Cockpit uses the shutdown command to shut down[src], but systemctl poweroff might work. I will also attempt to revert bioses if msi supports it. thank you very much!
@potentiallynotfelix Ok bet of luck. I’ve had all sorts of weird issues with systemd as of late, but not sure how many of those are inherent to systemd itself and how many are Ubuntu’s config of same.
sudo systemctl rebootdid the same. I’m starting to think this is bios related.@potentiallynotfelix Well flash to an older and see how it goes. I’ve seen some wired bios issues. I’ve got an i7=6850k machine on an Asus motherboard, and after I flashed to the latest bios, the USB power strobed on and off every few seconds so keyboard and mouse would work then not work then work then not work. I thought something was broken with hardware but then found others had the same issue with the most current BIOS, flashed to one release earlier and all good.
Is it actual server hardware? I’ve seen some very weird things with real servers that take ages to reboot (I was assuming it was self checking or something). Are you sure its hung, and not just very slow to shutdown/reboot?
Is there any serial/monitor output before the hang?
Monitor output after shutting down:
I’ve given it 6 hours or so to shut down, so it’s almost 100% a hang not a slow shutdown
I had this issue: failed to finalise remaining DM devices. Which led me to here https://github.com/systemd/systemd/issues/15004 and Skinner927 mentions your issue in that thread
I’d try uninstalling nouveau completely and see if the issue persists for you
The
xserver-xorg-video-nouveaupackage was not installed, how else would I remove nouveau?that’s only the X11 “driver” for it. nouveau is built into the kernel, the way to “uninstall” it is to make it not get loaded, by blacklisting it
https://wiki.archlinux.org/title/Nouveau
but this does not seem to be the problem
Agreed,
lsmod | grep nouveaureturns nothing, so I’m not concerned about nouveau or nvidia being the issue here.
I don’t have an Nvidia GPU so I don’t have any experience with it but a quick search brought me to Nvidias website and the instructions seem to line up with users answers on other forums.
Disable it here https://docs.nvidia.com/ai-enterprise/deployment/vmware/latest/nouveau.html or apparently installing Nvidias proprietary drivers automatically blacklists Nouveau.
lsmod | grep nouveaureturns nothing, so I assume removing my gpu automatically stopped it from being loaded. that sorta rules out nouveau as an issue assuming it’s not present somewhere else.
Direct link to skinners comment: https://github.com/systemd/systemd/issues/15004#issuecomment-2264687287
seems its a nvidia issue, i also have that issue, the gpu locks and i need to reboot while the VM with the nvidia passthrough freezes. i need a full reboot from baremetal machine to stop gpu using all his power stuck, don’t let it be for hours being on or you will kill your hardware
I have removed my gpu and the issue is still present.
yes sorry I read you after writting it, if you remove the GPU the log message is the same but without the GPU lockup line?
do you know that use device mapper? what kind of device is /dev/dm-1 ?
“dmsetup info” might help
sudo dmsetup inforeturns:Name: raven--vg-root State: ACTIVE Read Ahead: 256 Tables present: LIVE Open count: 1 Event number: 0 Major, minor: 254, 0 Number of targets: 1 UUID: LVM-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx Name: raven--vg-swap_1 State: ACTIVE Read Ahead: 256 Tables present: LIVE Open count: 2 Event number: 0 Major, minor: 254, 1 Number of targets: 1 UUID: LVM-xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxdid you make these yourself? if not, could you cdo an
ls -l /dev/mapper? it shows which name corresponds to which dm device
Says reboot, are you issuing a reboot or a shutdown poweroff? Entering sleep state 5 shout be power off right?
I click the reboot button on cockpit, which issues a
shutdown --rebootcommand as root. I agree that sleep state S5 is powered off. From the acpi docs:A computer state where the computer consumes a minimal amount of power. No user mode or system mode code is run. This state requires a large latency in order to return to the Working state. The system’s context will not be preserved by the hardware. The system must be restarted to return to the Working state. It is not safe to disassemble the machine in this state.
This likely means my system is failing to reach that s5/g2 state.
If you ssh login directly and issue same command, not In cockpit interface, does it react the same?
no, sorry for not specifying. it’s scrapped together from old consumer components.
- i3-10100
- 16gb ddr4 ram
- MSI h410m pro
halt -p
thanks for the suggestion, could you elaborate on what this would do differently from the regular shutdown command that systemctl uses? thanks again
My understanding is that ‘halt’ had been an alias for ‘halt -p’, but that changed recently. -p tells the command to power off. Without it, it just shuts down process.
halt -pdid nothing different. still hung on shutdown.
Your machine isn’t shutting down, it’s trying to sleep.
You also have active KVM instances which are fighting to keep it alive.
can you elaborate on why you suspect this? The cockpit reboot or shutdown button uses the
shutdowncommand directly along with a--rebootor--poweroffflag.onSubmit(event) { const Dialogs = this.context; const arg = this.props.shutdown ? "--poweroff" : "--reboot"; if (!this.props.shutdown) cockpit.hint("restart"); cockpit.spawn(["shutdown", arg, this.state.when, this.state.message], { superuser: "require", err: "message" }) .then(this.props.onClose || Dialogs.close) .catch(e => this.setState({ error: e.toString() })); event.preventDefault(); return false; }(source)
deleted by creator
