Unresponsive PC after x minutes of idle time

cstadegaart · 2011-07-21 21:33:04

Hello,

For my hobby I have a little server running Arch Linux. To practice a bit with Linux and have server running for different purposes. It boots without any input or output device, just a LAN connection. Works fine.

For some reason, after an unknown number of minutes the server is unresponsive when accessed via HTTP or SSH. I have to physically connect a USB keyboard and hit a button to let it become responsive again.

This is very annoying. I installed no specific packages to handle anything with power management. I installed acpid to handle my power button, but besides that, nothing.

I checked my BIOS settings for any power saving options and just to make sure it's not the BIOS configuration, I turned off any power saving option in there.

I just don't know where to look. In my frame of reference I looked everywhere.

I would be very glad to receive help and if I need to post some more information, please feel free to tell me what.

Thanks! Chris.

thoffmeyer · 2011-07-22 00:43:19

This could be your network card. Do you know what card it has?

cstadegaart · 2011-07-22 11:38:24

It is a D-Link DGE-528T.

See also this product information page.

Chris.

MoonSwan · 2011-07-22 19:35:59

I know from experience that some D-Link cards can be put to "sleep" by their own driver. I think the tool I used to find this out was called "ethtool" (not sure what app provides but pacman should help you there). With that app I could manipulate some of my Nic's functions such as power-saving being put to "on" or "off", IE with the Nic in "on/power-save = yes" my Nic would go to sleep when the computer was idle. I'm sorry I'm not much help but I hope this might push you in the right direction.

Oh there is also "mii" but it's old and outdated, it's a predecessor of Eth-tool IIRC.

Hope that helps!

cstadegaart · 2011-07-22 21:35:58

Yaourt did the job and installed ethtool for me.

Unfortunately I couldn't find any means to change the power management of the ethernet card... well not by addressing the man page of ethtool.

output of sudo lspci -v -s 00:08 > /mnt/nasserver/lspci_output:

00:08.0 Ethernet controller: D-Link System Inc DGE-530T Gigabit Ethernet Adapter (rev 11) (rev 11)
    Subsystem: D-Link System Inc DGE-530T Gigabit Ethernet Adapter (rev 11)
    Flags: bus master, 66MHz, medium devsel, latency 32, IRQ 16
    Memory at eb000000 (32-bit, non-prefetchable) [size=16K]
    I/O ports at d000 [size=256]
    [virtual] Expansion ROM at 40000000 [disabled] [size=128K]
    Capabilities: [48] Power Management version 2
    Capabilities: [50] Vital Product Data
    Kernel driver in use: skge
    Kernel modules: skge

It does have power management capabilities, so I'm going to check the skge driver page for more information myself later today (kind of working right now).

Thanks for the tips. If I find a solution, I'll let you know. If you have any hints, please feel free to post them!

Chris.

cstadegaart · 2011-07-23 15:12:31

Well this driver page didn't help me out much, since it's for a Solaris build.

I searched a lot of the net about Power Management, skge module and Arch Linux, but nothing helpful comes up.

I'll just keep on trying to find something.

Greets, Chris.

cstadegaart · 2011-07-25 20:17:09

Still nothing useful found.

lspci shows DGE-530T as the ethernet card (see earlier posts), so I'm starting to wonder which is right. My receipt from the reseller (which says DGE-528T) or lspci. Only one way to find out: open the PC and check the card itself.

Anyway, the 528T shouldn't use skge. 530T should though.

You never know if this is the issue. I will check back on the board ASAP.

Chris.

MoonSwan · 2011-07-25 20:35:39

Apparently there is still an "mii-tools" command in my system (no idea why) and it does seem to have a log function and a watch function to help debug things I'm guessing. This might be useful to you, I hope. It's the only thing I have right now. :\

Last edited by MoonSwan (2011-07-25 20:35:53)

cstadegaart · 2011-07-25 21:27:16

Maybe a process is running which handles power saving? If so, I guess it must be one started by kthreadd (kernel threads).

Still looking for a solution. Will open up the PC later today and thanks for the input MoonSwan, will check the log if I have the same prog.

ps -efH kcmd outputs:

UID        PID  PPID  C STIME TTY      STAT   TIME CMD
root         2     0  0 11:01 ?        S      0:00 [kthreadd]
root       126     2  0 11:01 ?        S<     0:00   [ata_sff]
root        12     2  0 11:01 ?        S      0:00   [bdi-default]
root       570     2  0 11:01 ?        S      0:09   [cifsd]
root         8     2  0 11:01 ?        S<     0:00   [cpuset]
root        20     2  0 11:01 ?        S<     0:00   [crypto]
root       152     2  0 11:01 ?        S<     0:00   [ext4-dio-unwrit]
root       323     2  0 11:01 ?        S<     0:00   [ext4-dio-unwrit]
root       512     2  0 11:01 ?        S      0:03   [flush-8:0]
root       932     2  0 12:39 ?        S      0:00   [flush-cifs-1]
root        19     2  0 11:01 ?        S      0:00   [fsnotify_mark]
root       151     2  0 11:01 ?        S      0:00   [jbd2/sda3-8]
root       322     2  0 11:01 ?        S      0:01   [jbd2/sda4-8]
root        13     2  0 11:01 ?        S<     0:00   [kblockd]
root         9     2  0 11:01 ?        S<     0:00   [khelper]
root       120     2  0 11:01 ?        S      0:00   [khubd]
root        18     2  0 11:01 ?        SN     0:01   [khugepaged]
root        15     2  0 11:01 ?        S      0:00   [khungtaskd]
root       296     2  0 11:01 ?        S<     0:00   [kpsmoused]
root        17     2  0 11:01 ?        SN     0:00   [ksmd]
root         3     2  0 11:01 ?        S      0:00   [ksoftirqd/0]
root        16     2  0 11:01 ?        S      0:15   [kswapd0]
root        22     2  0 11:01 ?        S<     0:00   [kthrotld]
root      1081     2  0 12:56 ?        S      0:00   [kworker/0:0]
root      1014     2  0 12:50 ?        S      0:00   [kworker/0:1]
root        23     2  0 11:01 ?        S      0:05   [kworker/0:2]
root         5     2  0 11:01 ?        S      0:00   [kworker/u:0]
root       129     2  0 11:01 ?        S      0:00   [kworker/u:1]
root         6     2  0 11:01 ?        S      0:00   [migration/0]
root        10     2  0 11:01 ?        S<     0:00   [netns]
root       127     2  0 11:01 ?        S      0:00   [scsi_eh_0]
root       128     2  0 11:01 ?        S      0:00   [scsi_eh_1]
root        11     2  0 11:01 ?        S      0:00   [sync_supers]
root         7     2  0 11:01 ?        S      0:00   [watchdog/0]
root         1     0  0 11:01 ?        Ss     0:00 init [3]  
root       627     1  0 11:01 ?        Ss     0:00   /usr/sbin/acpid
root       775     1  0 11:02 tty1     Ss+    0:00   /sbin/agetty -8 -s 38400 tty1 linux
root       776     1  0 11:02 tty2     Ss+    0:00   /sbin/agetty -8 -s 38400 tty2 linux
root       777     1  0 11:02 tty3     Ss+    0:00   /sbin/agetty -8 -s 38400 tty3 linux
root       778     1  0 11:02 tty4     Ss+    0:00   /sbin/agetty -8 -s 38400 tty4 linux
root       779     1  0 11:02 tty5     Ss+    0:00   /sbin/agetty -8 -s 38400 tty5 linux
root       780     1  0 11:02 tty6     Ss+    0:00   /sbin/agetty -8 -s 38400 tty6 linux
root       601     1  0 11:01 ?        Ss     0:00   /usr/sbin/crond -S -l info
root       527     1  0 11:01 ?        Ss     0:00   /sbin/dhcpcd -q eth0
nzbbox     687     1 17 11:01 ?        Sl    20:21   python /usr/share/sabnzbd/SABnzbd.py -d -f /home/nzbbox/.sabnzbd/sabnzbd.ini
nzbbox     756     1  0 11:02 ?        Sl     0:17   python /usr/share/sickbeard/SickBeard.py --daemon
root       992     1  0 12:43 ?        Ss     0:00   sshd: nzbbox [priv]
nzbbox     994   992  0 12:43 ?        S      0:00     sshd: nzbbox@pts/1
nzbbox     995   994  0 12:43 pts/1    Ss     0:00       -bash
nzbbox    1113   995  0 12:57 pts/1    R+     0:00         ps -efH kcmd
root       480     1  0 11:01 ?        S      0:00   supervising syslog-ng
root       481   480  0 11:01 ?        Ss     0:00     /usr/sbin/syslog-ng
root       205     1  0 11:01 ?        Ss     0:00   /sbin/udevd --daemon
root      1083   205  0 12:56 ?        S      0:00     /sbin/udevd --daemon
root      1084   205  0 12:56 ?        S      0:00     /sbin/udevd --daemon

MoonSwan · 2011-07-26 01:42:54

Stab in the dark follows: Try booting with acpi=off and apic=off? This might work because it is obviously sleeping because of "something."

cstadegaart · 2011-07-27 16:09:21

Well I finally opened up my PC and checked the NIC. It is a D-Link DGE-530T, so lspci is right (that's kind of a relief).

MoonSwan, I checked my system for this mii tool, but I guess is deprecated and replaced by ethtool.

Instead of booting with acpi or apic off I really wish to find the real source of the sleeping pill.

I went through the kernel documentation a bit to find anything useful there, but there are many things to try when you start reading, so I really could use someone experienced on this issue. Hopefully someone can help me by pointing me to the right direction.

Thanks MoonSwan for the input.

Chris.

MoonSwan · 2011-07-27 20:20:09

Yes mii-tool is deprecated, I happen to have it because I installed net-tools and haven't got round to uninstalling that package.

You're welcome, but I would have liked to have seen this issue solved. Sorry I wasn't much more help.

Leonid.I · 2011-07-27 21:19:11

dmesg.log should have info about the kernel module responsible for the nic. Then use modinfo <module_name> to find out if it is possible to disable power saving... if this is the cause.

I guess, booting with acpi=off is going to kill powersave system-wide, which might be an overkill.

Are you actually sure it's the NIC as opposed to something else? Have you tried to connect through wifi/another card?

cstadegaart · 2011-07-27 23:48:34

Already found out via lspci that skge is the kernel module which is responsible for the NIC and already did a modinfo on skge to find out if there are any parameters I can change, but unfortunately none can be which can help me out.

Nothing pops up when kernel boot options for increased debugging are enabled and the sleeping pill gets ignited. I wish some info messages would appear, but the source keeps itself hidden from me. Maybe extreme debugging would do the trick, but this will go heavy on the PC resources and thus my system would probably kill itself before it reaches the infamous sleep point.

Maybe you're right and I should have two NICs running and see if the other one still is responding. There is a disabled onboard NIC which can help me out here.

Will report back ASAP.

Chris.

Arch Linux

#1 2011-07-21 21:33:04

Unresponsive PC after x minutes of idle time

#2 2011-07-22 00:43:19

Re: Unresponsive PC after x minutes of idle time

#3 2011-07-22 11:38:24

Re: Unresponsive PC after x minutes of idle time

#4 2011-07-22 19:35:59

Re: Unresponsive PC after x minutes of idle time

#5 2011-07-22 21:35:58

Re: Unresponsive PC after x minutes of idle time

#6 2011-07-23 15:12:31

Re: Unresponsive PC after x minutes of idle time

#7 2011-07-25 20:17:09

Re: Unresponsive PC after x minutes of idle time

#8 2011-07-25 20:35:39

Re: Unresponsive PC after x minutes of idle time

#9 2011-07-25 21:27:16

Re: Unresponsive PC after x minutes of idle time

#10 2011-07-26 01:42:54

Re: Unresponsive PC after x minutes of idle time

#11 2011-07-27 16:09:21

Re: Unresponsive PC after x minutes of idle time

#12 2011-07-27 20:20:09

Re: Unresponsive PC after x minutes of idle time

#13 2011-07-27 21:19:11

Re: Unresponsive PC after x minutes of idle time

#14 2011-07-27 23:48:34

Re: Unresponsive PC after x minutes of idle time

Board footer