[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Readonly filesystem on boot



On Fri, Mar 02, 2012 at 12:13:31PM +0200, Bob Proulx wrote:
> Johann Spies wrote:
> > My home network server has started to give this problem some weeks ago:
> > When it boot, the root filesystem is mounted readonly.  
> 
> And it doesn't get remounted read-write?  

No.

> I suspect one of three problems.
> 
> To begin with the root filesystem is always booted read-only.

I am aware of that.

> If your system is not transitioning through from read-only to
> read-write then something in that process has been broken.  And there
> are many ways it may be broken so you will need to determine what has
> actually happened and fix it.

That is my objective :)

> I would check that you haven't accidentally removed the startup
> symlink for the /etc/rcS.d/S*mountall.sh which IIRC will be the
> startup process that mounts everything including the root filesystem.

I have renewed them also using rcconf.


> Check /var/log/syslog for any error messages that might be a clue.

Did that.  The problem with a readonly filesystem is that the boot
process cannot write to the logs :(


> Check the output of 'dmesg | less' for any information that might be a
> clue.

The only errors (more of a warning) I could see in dmesg was in
/var/log/dmesg and was related to the bios-bug I referred to in my
original message:

 0.000000] WARNING: BIOS bug: CPU MTRRs don't cover all of memory, losing 768MB of RAM.
[    0.000000] ------------[ cut here ]------------
[    0.000000] WARNING: at /build/buildd-linux-2.6_3.2.6-1-amd64-AkLxz2/linux-2.6-3.2.6/debian/build/source_amd64_none/arch/$
[    0.000000] Hardware name: MS-7528
[    0.000000] Modules linked in:
[    0.000000] Pid: 0, comm: swapper Not tainted 3.2.0-1-amd64 #1
[    0.000000] Call Trace:
[    0.000000]  [<ffffffff81046871>] ? warn_slowpath_common+0x78/0x8c
[    0.000000]  [<ffffffff816b35ff>] ? mtrr_trim_uncached_memory+0x2a8/0x2cd
[    0.000000]  [<ffffffff816ada9e>] ? setup_arch+0x4f1/0xb05
[    0.000000]  [<ffffffff8133a2b0>] ? printk+0x43/0x48
[    0.000000]  [<ffffffff81070747>] ? arch_local_irq_disable+0x7/0x8
[    0.000000]  [<ffffffff8107073f>] ? arch_local_irq_restore+0x7/0x8
[    0.000000]  [<ffffffff816a984f>] ? start_kernel+0xcf/0x3c8
[    0.000000]  [<ffffffff816a9140>] ? early_idt_handlers+0x140/0x140
[    0.000000]  [<ffffffff816a93c4>] ? x86_64_start_kernel+0x104/0x111
[    0.000000] ---[ end trace a7919e7f17c0a725 ]---
[    0.000000] update e820 for mtrr
[    0.000000] modified physical RAM map:

and some more memory related stuff.


> If that didn't yield anything then I would boot into single user mode
> and manually run each startup script up through mountall.sh and then
> debug running that script to see what the problem is.  I usually debug
> scripts by running them with 'sh -x scriptname' and looking at the
> trace output.

Thanks.  I will try that.
> 
> > Then I restart the network and go back to level 5.
> 
> You mean 2 sir.  The normal Debian runlevel is 2.  Runlevel 5 is used
> by some other distros.  But you are running Debian, right?  :-)
> Of course you could have edited /etc/inittab and changed the default.

You got me there.  I have worked previously as sysadm on RedHat, OpenSuse and
Debian for 10 years and forgot Debian use level 2 the way the others use
5. :)

> > All that did not make any difference.  There is a bios bug that causes
> > Linux not to see all 4G ram and the kernel is complaining about that
> > during the startup.
> 
> This is a completely different and separate issue.  Which kernel are
> you running?  'cat /proc/version'  Because due to limitations of the
> original pc architecture a standard 32-bit kernel won't be able to use
> all 4G.  If you are running a 32-bit kernel you will probably need the
> bigmem flavor.  If you want a flat address space you will need to run
> the amd64 kernel.  And again, the dmesg output will show the memory map.

I know about the limitation of the 32-bit kernel.  Before the amd64's
was available I would use a PAE-kernel and I have indeed tried that also
with this computer when it was still new and I could not get success
with the amd64-kernels.  But I had now luck.

I have quoted the dmesg-output about this earlier.

Thanks for your trouble to help me think.  

Regards
Johann
-- 
Johann Spies                            Telefoon: 021-808 4699
Databestuurder /  Data manager

Sentrum vir Navorsing oor Evaluasie, Wetenskap en Tegnologie
Centre for Research on Evaluation, Science and Technology 
Universiteit Stellenbosch.

     "I will lift up mine eyes unto the hills, from whence 
      cometh my help. My help cometh from the LORD, which 
      made heaven and earth."                      
                           Psalms 121:1,2 


Reply to: