[gentoo-user] Magic SysRq didn't work today

[gentoo-user] Magic SysRq didn't work today

[Mark Knecht]

Hi,
   For the first time in a couple of years I had a total hard hang
today when a Virtualbox VM was doing some updates to M$ Office. The
compute server was completely alive & then completely hung. The mouse
wouldn't move, I couldn't switch to the console and I couldn't log in
from another machine. I tried Alt-Ctrl-SysRq REISUB but there was no
response to that either so I went with a hard reset and did some
fsck's to ensure everything was OK. (It was...) After bringing the
machine back up I tested that the Magic SysRq worked, and it did, so
the hang was hard enough to stop even that.

   My question about this is what's your guess about where the hang
occurred? Was it purely in the kernel and I just hit some corner
condition that caused it? Or could it be that a hang in the VM could
completely bring the machine to a halt such that the kernel never
gains control even after waiting 15 minutes?

   The kernel was gentoo-sources-3.0.4, not the -r1 that came out today.

   Looking forward to all ideas.

Cheers,
Mark

Re: [gentoo-user] Magic SysRq didn't work today

[Dale]

Mark Knecht wrote:
> Hi,
>     For the first time in a couple of years I had a total hard hang
> today when a Virtualbox VM was doing some updates to M$ Office. The
> compute server was completely alive&  then completely hung. The mouse
> wouldn't move, I couldn't switch to the console and I couldn't log in
> from another machine. I tried Alt-Ctrl-SysRq REISUB but there was no
> response to that either so I went with a hard reset and did some
> fsck's to ensure everything was OK. (It was...) After bringing the
> machine back up I tested that the Magic SysRq worked, and it did, so
> the hang was hard enough to stop even that.
>
>     My question about this is what's your guess about where the hang
> occurred? Was it purely in the kernel and I just hit some corner
> condition that caused it? Or could it be that a hang in the VM could
> completely bring the machine to a halt such that the kernel never
> gains control even after waiting 15 minutes?
>
>     The kernel was gentoo-sources-3.0.4, not the -r1 that came out today.
>
>     Looking forward to all ideas.
>
> Cheers,
> Mark
>
>

Well, I had a weird one a while back that caused this sort of issue.  It 
turned out that one of my hard drives was having a file system issue.  I 
put a new file system on it, it works fine ever since.  Weird but real.

That said, if the kernel itself locks up, the SysReq keys may not work.  
I have found that if you can't ssh in or get the SysReq keys to work, it 
is what I call a hard lock up.  Hitting the reset button or pulling the 
plug is the only option I know of.

You can set it so that it will reboot automatically when it locks up.  
Thing is, if the lock up is severe enough, that may not work either.  
Same as the SysReq keys I guess.

Having the option to use SysReq keys are nice but sometimes even that 
doesn't work.  It would be nice if one could cross their fingers when 
using those.  I have to use both hands to press mine.  Does crossing 
toes help?  lol

Dale

:-)  :-)

Re: [gentoo-user] Magic SysRq didn't work today

[Pandu Poluan]

[-- Attachment #1: Type: text/plain, Size: 2236 bytes --]

On Sep 29, 2011 6:55 AM, "Dale" <rdalek1967@gmail.com> wrote:
>
> Mark Knecht wrote:
>>
>> Hi,
>>    For the first time in a couple of years I had a total hard hang
>> today when a Virtualbox VM was doing some updates to M$ Office. The
>> compute server was completely alive&  then completely hung. The mouse
>> wouldn't move, I couldn't switch to the console and I couldn't log in
>> from another machine. I tried Alt-Ctrl-SysRq REISUB but there was no
>> response to that either so I went with a hard reset and did some
>> fsck's to ensure everything was OK. (It was...) After bringing the
>> machine back up I tested that the Magic SysRq worked, and it did, so
>> the hang was hard enough to stop even that.
>>
>>    My question about this is what's your guess about where the hang
>> occurred? Was it purely in the kernel and I just hit some corner
>> condition that caused it? Or could it be that a hang in the VM could
>> completely bring the machine to a halt such that the kernel never
>> gains control even after waiting 15 minutes?
>>
>>    The kernel was gentoo-sources-3.0.4, not the -r1 that came out today.
>>
>>    Looking forward to all ideas.
>>
>> Cheers,
>> Mark
>>
>>
>
> Well, I had a weird one a while back that caused this sort of issue.  It
turned out that one of my hard drives was having a file system issue.  I put
a new file system on it, it works fine ever since.  Weird but real.
>
> That said, if the kernel itself locks up, the SysReq keys may not work.  I
have found that if you can't ssh in or get the SysReq keys to work, it is
what I call a hard lock up.  Hitting the reset button or pulling the plug is
the only option I know of.
>
> You can set it so that it will reboot automatically when it locks up.
 Thing is, if the lock up is severe enough, that may not work either.  Same
as the SysReq keys I guess.
>
> Having the option to use SysReq keys are nice but sometimes even that
doesn't work.  It would be nice if one could cross their fingers when using
those.  I have to use both hands to press mine.  Does crossing toes help?
 lol

Naaah. It *must* be the fingers of your hands. Employ your little finger and
thumb, freeing the index and middle to do the all-important crossing...

;-)

Rgds,

[-- Attachment #2: Type: text/html, Size: 2765 bytes --]

Re: [gentoo-user] Magic SysRq didn't work today

[Spidey / Claudio]

I have lived through some lock ups in the recent past, but that's
because I've disassembled my desktop from it's case and assembled it
at my working table. Since both PS/2 ports of the mobo are on my mouse
pad (yeah, short cables, tight space), I eventually pull some cable os
slap my video card. The first time X.org locked up, but ssh'ing from
my Maemo phone did the trick (had to disable kexec and reboot from the
bios, though). The second time I couldn't do that, my video card
cooler fans stopped, and I had to pull the power plug and wait a few
moments. That time I thought that some money would be spent on a new
motherboard or video card. But rest assured, no computer components
were harmed in those accidents.

Ending the off topic, I had lock downs and headaches when messing with
reiser4. I'd say that recently built experimental code in the kernel
is something that would trigger lock ups.

Re: [gentoo-user] Magic SysRq didn't work today

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 1337 bytes --]

On Thursday 29 Sep 2011 04:40:15 Spidey / Claudio wrote:
> I have lived through some lock ups in the recent past, but that's
> because I've disassembled my desktop from it's case and assembled it
> at my working table. Since both PS/2 ports of the mobo are on my mouse
> pad (yeah, short cables, tight space), I eventually pull some cable os
> slap my video card. The first time X.org locked up, but ssh'ing from
> my Maemo phone did the trick (had to disable kexec and reboot from the
> bios, though). The second time I couldn't do that, my video card
> cooler fans stopped, and I had to pull the power plug and wait a few
> moments. That time I thought that some money would be spent on a new
> motherboard or video card. But rest assured, no computer components
> were harmed in those accidents.
> 
> Ending the off topic, I had lock downs and headaches when messing with
> reiser4. I'd say that recently built experimental code in the kernel
> is something that would trigger lock ups.

I was running reiser4 for more than a year and found it rather temperamental 
on my hardware (hard lock ups, corrupted fs, etc).  Eventually, I replaced it 
with ext4 and have not had problems since.  For me reiser4 seemed to perform 
better (as in faster) than ext4, in all but mounting speeds.

YMMV
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Magic SysRq didn't work today

[Spidey / Claudio]

Deleting files were slow as hell too.

Claudio Roberto França Pereira (a.k.a. Spidey)
hardMOB - HTForum - @spideybr
Engenharia de Computação - UFES 2006/1



On Thu, Sep 29, 2011 at 08:39, Mick <michaelkintzios@gmail.com> wrote:
> On Thursday 29 Sep 2011 04:40:15 Spidey / Claudio wrote:
>> I have lived through some lock ups in the recent past, but that's
>> because I've disassembled my desktop from it's case and assembled it
>> at my working table. Since both PS/2 ports of the mobo are on my mouse
>> pad (yeah, short cables, tight space), I eventually pull some cable os
>> slap my video card. The first time X.org locked up, but ssh'ing from
>> my Maemo phone did the trick (had to disable kexec and reboot from the
>> bios, though). The second time I couldn't do that, my video card
>> cooler fans stopped, and I had to pull the power plug and wait a few
>> moments. That time I thought that some money would be spent on a new
>> motherboard or video card. But rest assured, no computer components
>> were harmed in those accidents.
>>
>> Ending the off topic, I had lock downs and headaches when messing with
>> reiser4. I'd say that recently built experimental code in the kernel
>> is something that would trigger lock ups.
>
> I was running reiser4 for more than a year and found it rather temperamental
> on my hardware (hard lock ups, corrupted fs, etc).  Eventually, I replaced it
> with ext4 and have not had problems since.  For me reiser4 seemed to perform
> better (as in faster) than ext4, in all but mounting speeds.
>
> YMMV
> --
> Regards,
> Mick
>

Re: [gentoo-user] Magic SysRq didn't work today

[Mick]

[-- Attachment #1: Type: Text/Plain, Size: 850 bytes --]

On Thursday 29 Sep 2011 18:35:13 Spidey / Claudio wrote:
> Deleting files were slow as hell too.

Hmm ... no, not here.  Reiser4 pretty much blew the doors off anything else I 
have ever used in Linux land.  (I have not tried btrfs yet).

I can't even blame reiser4 for the 'temperamental' behaviour that I mentioned 
earlier with any degree of certainty.  Some people have blamed kernel patches 
that borked reiser4, so heed Volker's earlier comment.  I have also found some 
gcc versions not being particularly friendly to it.

The only repetitive failure modes I noticed where I could definitely blame the 
fs were:

a) recovery after power failure (unlike e.g. good ol' reiserfs)
b) running out of space on a partition.

Once these two failure modes are guarded against the fs was performing as 
expected.
-- 
Regards,
Mick

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 198 bytes --]

Re: [gentoo-user] Magic SysRq didn't work today

[Niccolò Belli]

Il 29/09/2011 00:04, Mark Knecht ha scritto:
>     For the first time in a couple of years I had a total hard hang

You are lucky, it happened tons of times to me.