Openbox re-starting PC?

[Martin1978]

Hi everyone!

Firstly, thanks to Melodie and anyone else involved in producing the Openbox PCLOS. I just installed installed it today and the first impressions are very favourable 

Having a small problem however when it comes to shutting down - the pc will turn off for 10-20 seconds and then reboot. I think I saw something about this issue in another thread, but I cannot find it now. Is there a solution to this? Please bear in mind that I am fairly new to linux and so I might need a little extra help if the solution is tricky!

Thanks

Martin

[ka9yhd]

Welcome. 

What kind of PC are you running Openbox on?

[nightcap]

pc will turn off for 10-20 seconds and then reboot.

Seems I've been through a similar bizarre situation recently with my HP ProBook 6450b (i3-370). Installed PCLinuxOS OpenBox and all was well for a couple of weeks. Then as you've described: involuntary boot ups several seconds following clean shut down.  Also it appeared to intermittently disable my external power adaptor.  

I do not claim to (a) know the root cause or (b) have a universally-applicable answer (many variables - hardware/BIOS/install setup/etc). FWIW I checked exhaustively in BIOS and found a setting had been changed without my knowing. "Wake By USB" was suddenly switched on.    Turned that off immediately.

Now here's another intriguing aspect. Following that BIOS fix, after shutting down a number of times, there was little change - still involuntary boot ups. Left it overnight. Came back. Voila! No more issues. Being a curious cat I decided to install another distro altogether to see if I could replicate original issues, with "Wake By USB" BIOS setting switched on and off. Have not seen any issues with other distro. So I'm still testing and awaiting further findings. Trust this helps (a little). YMMV.

nightcap

[melodie]

Hi,

Try the option "acpi=off" on the kernel command line. It has solved it a few times for this issue (which happens only with Openbox versions, and on some machines only).

Regards,
Mélodie

[nightcap]

Thanks Melodie. All my linux installs are with ACPI switched off in BIOS. So your advice regarding kernel setting makes perfect sense.

OK, further info. My linux install following PCLOS OpenBox is (dare I mention it in this forum?) Archbang, which runs OpenBox 3.5. So far, it appears to exhibit no issues with involuntary boot ups. Will monitor and report back.

nightcap

[melodie]

Hi,
What kernel does the other distro use ? And if the kernel is the same number (I'd be astonished) what are the differences in the settings of the kernel ? (you can compare the files /proc/config.gz).

The former kernel didn't produce this effect : the 2.6.37 which was in use before didn't. We all had a very long thread about that, when the Bonsai version was in test this summer. I searched a long time, did compare with the content (packages installed) in the other PCLinuxOS versions (Zen Mini, KDE Minime kernel, LXDE Mini) it has taken a lot of my time and didn't find anything clear, particularly as it does it only on some machines. I have solved it for a while by adding "/sbin/halt;" at the beginning of the GDM shutdown line (you can configure GDM in the advanced section at login screen), this helped some time but the problem came back, in my P4 Dell with hyperthreading, whereas it seemed to have solved it for someone else, at the forum. I finally added "acpi=off" and since I didn't meet with the issue anymore.

Why have you turned the acpi off in the bios in all your machines ?

[Martin1978]

What kind of PC are you running Openbox on?

The computer is an HP Pavillion (AMD dual core). Thanks for your input nightcap and melodie,  I'll have a little play about and see what happens. Of course though, it's more of a puzzler than a problem.


Martin

[nightcap]

What kernel does the other distro use ? ------- what are the differences in the settings of the kernel ? (you can compare the files /proc/config.gz).

Why have you turned the acpi off in the bios in all your machines ?

Other kernel is 3.0 Arch. Will get back to you on kernel settings comparison when I have more time. Most linux distro forums recommend ACPI be switched off in BIOS. My limited understanding of ACPI is that it is a narrow protocol implemented in different ways by various hardware manufacturers and can be notoriously inconsistent in how it affects the stability of OSes. ACPI BIOS config tables are loaded each boot if allowed to do so. (BTW main ACPI development players are M$ and Intel.)

Several years back a brand new AMD-powered Compaq notebook I encountered produced BSODs ex-factory (M$ XP) for no apparent reason. Retailer reloaded OS three times (charging client each time). Same outcome each time - BSOD. Client and I researched the issue on-line and it appeared that both HP and M$ admitted a chronic problem, likely due to ACPI conflicts, but chose to evade the issue (for a low-volume, entry-level mainboard, AMD CPU and chipset, at the tail-end of XP tech support). Anyway I loaded linux with ACPI disabled in BIOS. That notebook has been rock solid ever since.

nightcap

[melodie]

I have been told, and read that acpi are (deliberately) buggy, especially for Linux based distributions. I have not done this test myself, but would be curious to give it a try one of these days.

[CaptainSarcastic]

I started getting similar behavior on my machine that runs PCLOS Gnome - a good portion of the time doing a clean shutdown results in an immediate attempt at rebooting, but not 100% of the time.  I've been working under the assumption it was my hardware doing it (there are some quirks there...) but I will check ACPI and might actually do some more troubleshooting now.

FWIW the motherboard is an ASUS running a socket 939 Athlon 64 x2.

[melodie]

FWIW the motherboard is an ASUS running a socket 939 Athlon 64 x2.

Hi,
It seems that many brands of motherboards would be constructed by the same manufacture (Foxconn). See the link above...

[nightcap]

I have been told, and read that acpi are (deliberately) buggy, especially for Linux based distributions. I have not done this test myself, but would be curious to give it a try one of these days.

Suggest there is minimal purpose served by grappling with ACPI, if one is fine-tuning linux, unless there is even-handed ACPI development support for linux. With M$ manipulating their "industry collaborators" to conform to their usual restrictive trade practices what hope is there for a level playing field?   

It seems that many brands of motherboards would be constructed by the same manufacture (Foxconn). See the link above...

Sorry, this would seem irrelevant, Melodie. Yes, Foxconn is a huge OEM Chinese manufacturer (>400,00 workers at their largest facility, the one where production line accidents, toxic poisoning and worker suicides have been reported). Each mainboard brand fashions its own style of BIOS, irrespective of where the board is assembled. For example, ASUS, Gigabyte, MSI, Intel, AsRock, HP, Lenovo, Toshiba, Fujitsu etc etc all differ. Then one factors in chipset type, overclocking/unlocking options, power management etc etc and many other variables. I'm certain others know a whole lot more about this.

nightcap

[Hootiegibbon]

I have been told, and read that acpi are (deliberately) buggy, especially for Linux based distributions. I have not done this test myself, but would be curious to give it a try one of these days.

Melodie,

The foxconn issue is fairly well documented, however it is not acpi in itself that is buggy, but the hardware implementation on certain boards.

This restart issue is specific to the openbox iso's all other PCLinuxOS versions remain (to my knowledge) unaffected, I suspect that there is a common-missing lib that is causing this that may or may not be acpi centric, it also makes sence (if these reboots are happening on machines with wake-on-lan active) to see if there is a missing lib in that regard as if the router on those networks is pining the machine - that would be enough to wake it (this would only apply if all occurrences happen on ethernet wired machines and none happened on wireless only)

I would suggest checking the similarity of the other versions - which libs are common to all versions, but missing in the openbox versions this would narrow down any further troubleshooting

Jase   

[melodie]

This restart issue is specific to the openbox iso's all other PCLinuxOS versions remain (to my knowledge) unaffected

Hi Hootie,

This is what I thought, until I read here : http://www.pclinuxos.com/forum/index.php/topic,99016.msg840949.html#msg840949

I suspect that there is a common-missing lib that is causing this that may or may not be acpi centric

Do as I did, take a small version (anyone), do a "rpm -qa" in it, do the same in Bonsai (which all Openbox versions come from since the 2nd edition), then on the result of each file, get a "sort" to have the files ordered by name, then a diff with the options "-aur" to get the order, the color and everything that makes it easier to compare.

Compare and tell me if you find something, because I didn't.

it also makes sence (if these reboots are happening on machines with wake-on-lan active) to see if there is a missing lib in that regard as if the router on those networks is pining the machine - that would be enough to wake it (this would only apply if all occurrences happen on ethernet wired machines and none happened on wireless only)

As far as I know, in the only machine where I also met with the issue, wake on lan was not enabled in the BIOS.

I would suggest checking the similarity of the other versions - which libs are common to all versions, but missing in the openbox versions this would narrow down any further troubleshooting

I have done that already, but feel free to do it too and report if you find anything possibly useful. Please take note that on a test Openbox version I also copied the whole kernel from the KDE version, including img, /lib/modules and so on, and also the halt initscript... and didn't get a different result.

Regards,
Mélodie

[Hootiegibbon]

I suspect that there is a common-missing lib that is causing this that may or may not be acpi centric

Do as I did, take a small version (anyone), do a "rpm -qa" in it, do the same in Bonsai (which all Openbox versions come from since the 2nd edition), then on the result of each file, get a "sort" to have the files ordered by name, then a diff with the options "-aur" to get the order, the color and everything that makes it easier to compare.

Compare and tell me if you find something, because I didn't.

Melodie,

I wont be troubleshooting this any further, as you are aware of my thoughts on the bonsai iso, IF there is no missing lib then it is a configuration option you have changed from the main releases - this is a long standing bug within your iso's and should have been dealt with pre-release.

it also makes sence (if these reboots are happening on machines with wake-on-lan active) to see if there is a missing lib in that regard as if the router on those networks is pining the machine - that would be enough to wake it (this would only apply if all occurrences happen on ethernet wired machines and none happened on wireless only)

As far as I know, in the only machine where I also met with the issue, wake on lan was not enabled in the BIOS.

I would suggest checking the similarity of the other versions - which libs are common to all versions, but missing in the openbox versions this would narrow down any further troubleshooting

I have done that already, but feel free to do it too and report if you find anything possibly useful. Please take note that on a test Openbox version I also copied the whole kernel from the KDE version, including img, /lib/modules and so on, and also the halt initscript... and didn't get a different result.

Regards,
Mélodie

I would suggest you check out any settings and modules related to PXE as there is potential that an incorrect variable  could restart a PC after shutdown.

I wish you luck in YOUR bug hunting

Jase