[linux-lvm] processes remain in D-state,, using lvm, reiserfs and VIA-Board kernel 2.4.2+

[linux-lvm] processes remain in D-state,, using lvm, reiserfs and VIA-Board kernel 2.4.2+

[Jan Walzer]

Hello people,

<short story>
using LVM and reiserfs on an disk connected on an VIA-Promise
controler, leaves different processes in D-state when accessing some
directories on this disk.
This is since kernel 2.4.2 until now on kernel 2.4.4-ac1
The disk then is 'busy' and can't be umounted, so there is no clean
way to reboot the system. sometimes even SysReq-S,SysReq-U won't
umount, stopping on the LVM-Device.
</short story>

<long story>
I don't know where to start.
My company gave me the job to care for the server in our new office. I
was told only to look for it a bit. They already had it configured.

It's an Athlon1,2GHz on an ASUS board with VIA and an Promise-IDE
controller. Filesystem was on an IBM 30GB-SCSI disk, connected to an
Adaptec. 1024MB-ram

The "best"-thing: One BIG '/'-partition, where nearly all the data is
stored on ... yeah these guys were good.
As they already made it ready, they rsynced the data of their server
in the old office with this one. so there are 'bout 20GB data on the
disk, meaning only having 6GB left.

No Problems so far, except for power failures, when it needs to fsck
the whole disk.

3weeks ago I got an 40MB-Maxtor-IDE for moving parts of the data on
it. I connected it to the Promise that is onboard. I decided, to do
it better than them, and create an LVM. I ended in having 2 LVs on
the disk, using Reiser. All seems to work well, except, for the told
problem.

</long story>

Is there a probleme like this known?
What further informations do you need ? - I'll try to support as much
information, as I can.

  

-- 

 ,',    Jan Walzer      \V/  http://wa.lzer.net     ,',
',','   student of      >|<  mailto:jan.w@lzer.net ',','
  '   ComputerScience   /A\  +49-177-7403863         '

Re: [linux-lvm] processes remain in D-state,, using lvm, reiserfs and VIA-Board kernel 2.4.2+

[Glenn Shannon]

Jan Walzer wrote:

> Hello people,
>
> <short story>
> using LVM and reiserfs on an disk connected on an VIA-Promise
> controler, leaves different processes in D-state when accessing some
> directories on this disk.
> This is since kernel 2.4.2 until now on kernel 2.4.4-ac1
> The disk then is 'busy' and can't be umounted, so there is no clean
> way to reboot the system. sometimes even SysReq-S,SysReq-U won't
> umount, stopping on the LVM-Device.
> </short story>
>
> <long story>
> I don't know where to start.
> My company gave me the job to care for the server in our new office. I
> was told only to look for it a bit. They already had it configured.
>
> It's an Athlon1,2GHz on an ASUS board with VIA and an Promise-IDE
> controller. Filesystem was on an IBM 30GB-SCSI disk, connected to an
> Adaptec. 1024MB-ram
>
> The "best"-thing: One BIG '/'-partition, where nearly all the data is
> stored on ... yeah these guys were good.
> As they already made it ready, they rsynced the data of their server
> in the old office with this one. so there are 'bout 20GB data on the
> disk, meaning only having 6GB left.
>
> No Problems so far, except for power failures, when it needs to fsck
> the whole disk.
>
> 3weeks ago I got an 40MB-Maxtor-IDE for moving parts of the data on
> it. I connected it to the Promise that is onboard. I decided, to do
> it better than them, and create an LVM. I ended in having 2 LVs on
> the disk, using Reiser. All seems to work well, except, for the told
> problem.
>
> </long story>
>
> Is there a probleme like this known?
> What further informations do you need ? - I'll try to support as much
> information, as I can.
>
>
>
> --
>
>  ,',    Jan Walzer      \V/  http://wa.lzer.net     ,',
> ',','   student of      >|<  mailto:jan.w@lzer.net ',','
>   '   ComputerScience   /A\  +49-177-7403863         '
>
> _______________________________________________
> linux-lvm mailing list
> linux-lvm@sistina.com
> http://lists.sistina.com/mailman/listinfo/linux-lvm

Here is a way to check (as I have had this problem before):

right before the line in /etc/rc.d/init.d/halt that states:
mount -n -o remount,ro /

put:
lsof >/root/lsofoutput.txt

Then reboot and check the lsofoutput.txt file and see what is being
used......my thinking is that you'll see a bunch of stuff in the /usr/lib
directory being used by various processes. That would indicate a problem
with the method used to copy the files to their new partitions. Meaning
that a bunch of the symlinks are now broken. I have scripts to repair all
of this so if you need, ask.

Let us know how the output from that goes.

TTYL

Glenn