Boot from S3 state failing

Hi forum!

It's been a while since I posted something, in part because my OMV has been working flawlessly. However, I am having some issues currently with rtcwake.

My server goes in S3 suspend mode daily via a cronjob using rtcwake, the reason being that I am out of home the whole day and keeping it running cost me 120€ of electricity a year. Shutting it down during the day cuts the running costs to ~30€/year. However, since I moved the OMV installation to a new rig I started having issues. The system will suspend as expected, but it will randomly fail to properly wake up. The system wakes up, but it does not boot back online. The system becomes completely unresponsive and I have to pull the plug of the server to hard-reset it.

The first thing I checked was the logs, but there is nothing in there. I can clearly see when the system goes into suspension, but there is nothing else after that. Therefore, I have some questions to try to narrow the problem:

1. I have only 2 GB of RAM and no SWAP (I am using an SSD as OS drive and have the flash memory plugin installed). The ram usage is usually around 30% and the OMV installation size is ~12,5 GB. Might it be that I have not enough ram to use suspension?

2. I moved the OS hard drive from my old rig to the new one. The problem was not present in the old one. The main difference is that in the old one I had a swap partition (I was not using the flash memory plugin). Then, might it be that the lack of swap might be the source of my problem?

Thanks for the help!

I notice the problem when I come back home and try to use Plex. Then I check other plugins and none work. I am guessing that the problem happens when the system goes online (or tries to) because the logs are not registering any system startup or wake up. The cronjob wakes the server at 4 pm everyday, but when the failure happens nothing is registered in the logs, leading me to think that the server wakes up but somehow OMV fails to boot.

I've not connected a monitor because I can't reproduce the error myself, it happens randomly. I know that none of OMV plugins or services run (everything from nginx to Plex stops working. All plugins are down). On top of that, the power button does not respond either.

Sent from my Nexus 7 using Tapatalk

@subzero79

I am ysing backports kernel (3.16 i think). Will try rolling back to the normal one.

@BitHoarder
No, the interface is also deade. No plugin responds. SSH is down, nginx (therefore no web interface), Plex, Samba, Owncloud, etc. Nothing runs when the problem happens. The system wakes up from suspension (the computer is on an running) but the OS does not boot back.

I reenabled the swap and set the system to hibernate instead. Will see if the failure happens. After work i will check also de kernel and roll it back to the standard one.

Sent from my LG-H955 using Tapatalk

@tinh_x7

I have to ask: have you explicitly configured your system to hibernate (suspend to disk) without having a swap partition? In my case, I had to re-enable the swap in order to be able to hibernate. I know that it is possible to achieve S4 without swap, but doing so requires another procedure. In your case the issue might be a hardware limitaion because AFAIK S5 (power off, rtcwake -m off) is not an official S-state and not all motherboards support it (in my case, for example). Are you sure that rtcwake shutdown works? From what I've tested in my rigs and from the rtcwake documentation, shutdown is not a recognized mode. Is the eth0 hanging upon shutdown only or also when going into S3 or S4 states?

@subzero79

The system is installed in a SSD and I installed the flash memory plugin to reduce the amount of writes to it. I doubt that the SSD is damaged, I checked the smart values and everything seems OK. As you recommended, I disabled the plug in and re-enabled the swap partition. I changed the suspension mode from S3 to S4 (hibernate) to see if that helps.

As tinh_x7 said, does enabling the swap defeats the purpose of the flash memory plugin?

What command are you using to shut your system down? So, the NIC hangs before the computer shuts down? Does it also happen if you manually shut down the server (or if you send poweroff thru CLI)?

00 22 0 3 root .... rctwake -m disk - s 34000

This does not shut down the system, it makes it hibernate. You can't hibernate without having a swap partition. If you enable swap and run the command again, chances are that the problem will be gone.

If you want to shut down the computer using rtcwkake, you need to use:

rtcwake -m off -s 34000

From the rtcwake documentation: "(off is) Not officially supported by ACPI, but usually working". off is not a real ACPI power state, and it depends on which motherboard/bios you have. Some support it, some don't. My computer, for example, does not support it. Using rtcwake off in my systems prompts me to use shutdown instead, because off is not supported by my motherboard/bios:

rtcwake: wakeup from "off" using /dev/rtc0 at Fri May 20 02:12:20 2016
shutdown: -H and -P flags can only be used along with -h flag.
Usage:    shutdown [-akrhPHfFnc] [-t sec] time [warning message]
                  -a:      use /etc/shutdown.allow
                  -k:      don't really shutdown, only warn.
                  -r:      reboot after shutdown.
                  -h:      halt after shutdown.
                  -P:      halt action is to turn off power.
                  -H:      halt action is to just halt.
                  -f:      do a 'fast' reboot (skip fsck).
                  -F:      Force fsck on reboot.
                  -n:      do not go through "init" but go down real fast.
                  -c:      cancel a running shutdown.
                  -t secs: delay between warning and kill signal.
                  ** the "time" argument is mandatory! (try "now") **

There you go! If you want to shut down, you have to use another command. Try enabling swap and running the cron job. Perhaps that solves your issue.

Zitat von tinh_x7

Have you try the work around method yet?
wiki.debian.org/Hibernation/Hibernate_Without_Swap_Partition

Nope, didn't had the time yet. I want to see first if the problem I am having happens also with hibernation.