2001-11-22 13:05:32

by Marcus Grando

[permalink] [raw]
Subject: Input/output error

Hi list,

After reboot in kernel 2.4.15-pre9 this problem occur:

On try start syslog deamon occur this errrors "Input/output error" on many archives /var directory.

I mount the /var at boot linux single, and when try ls on /var/run list all archives .pid,
before try ls -la on /var/run, not list all archives.
Try remove .pid archives and the error occur again.

/var partition use ext2

Any suggestion?

Tanks in advance,

Regards
Marcus Grando




2001-11-22 13:17:33

by Rasmus Bøg Hansen

[permalink] [raw]
Subject: Re: Input/output error

On Thu, 22 Nov 2001, Marcus Grando wrote:

> On try start syslog deamon occur this errrors "Input/output error" on many archives /var directory.

Try to run fsck on the /var partition. Also you should check the disk
for bad blocks. What output do you get from the kernel ('dmesg',
/var/log/messages etc.)?

It could be a bad disk developing bad sectors.

Rasmus

--
-- [ Rasmus 'M?ffe' B?g Hansen ] ---------------------------------------
DISCLAIMER: Microsoft, Windows, Windows 98, Bugs, Lacking features, IRQ
conflicts, System crashes, Non-functional multitasking, the Y2K problem
and the Blue Screen of Death are registered trademarks of
Microsoft, Corp., Redmond, USA.
--------------------------------- [ moffe at amagerkollegiet dot dk ] --

2001-11-23 15:43:52

by Hartmut Holz

[permalink] [raw]
Subject: Re: Input/output error

Rasmus B?g Hansen wrote:

> On Thu, 22 Nov 2001, Marcus Grando wrote:
>
>
>>On try start syslog deamon occur this errrors "Input/output error" on many archives /var directory.
>>
>
> Try to run fsck on the /var partition. Also you should check the disk
> for bad blocks. What output do you get from the kernel ('dmesg',
> /var/log/messages etc.)?
>
> It could be a bad disk developing bad sectors.
>
> Rasmus
>
>

On my machine (2.4.15 final) it is the same behaviour. After reboot
the lock files (and only the lock files) are corrupt. With 2.4.14 and
2.4.13 everything works fine. gcc 2.96, e2fsck 1.25, aic7896/97

e2fsck output:
--------------
Pass 1: Checking inodes, blocks, and sizes
Pass 2: Checking directory structure
Entry 'kudzu' in /lock/subsys (30001) has deleted/unused inode 30005.
Clear? yes

Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
Clear? yes

Entry 'syslog' in /lock/subsys (30001) has deleted/unused inode 30007.
Clear? yes

Entry 'portmap' in /lock/subsys (30001) has deleted/unused inode 30008.
Clear? yes

Entry 'nfslock' in /lock/subsys (30001) has deleted/unused inode 30009.
Clear? yes

Entry 'random' in /lock/subsys (30001) has deleted/unused inode 30012.
Clear? yes

Entry 'netfs' in /lock/subsys (30001) has deleted/unused inode 30013.
Clear? yes

Entry 'autofs' in /lock/subsys (30001) has deleted/unused inode 30014.
Clear? yes

Entry 'local' in /lock/subsys (30001) has deleted/unused inode 30029.
Clear? yes

Entry 'syslogd.pid' in /run (38001) has deleted/unused inode 38009.
Clear? yes

Entry 'klogd.pid' in /run (38001) has deleted/unused inode 38010.
Clear? yes

Pass 3: Checking directory connectivity
Pass 4: Checking reference counts
Pass 5: Checking group summary information

/var: ***** FILE SYSTEM WAS MODIFIED *****
/var: 23017/66000 files (5.9% non-contiguous), 115364/263168 blocks


2001-11-23 18:30:18

by Andreas Dilger

[permalink] [raw]
Subject: Re: Input/output error

On Nov 23, 2001 16:43 +0100, Hartmut Holz wrote:
> On my machine (2.4.15 final) it is the same behaviour. After reboot
> the lock files (and only the lock files) are corrupt. With 2.4.14 and
> 2.4.13 everything works fine. gcc 2.96, e2fsck 1.25, aic7896/97
>
> e2fsck output:
> --------------
> Pass 1: Checking inodes, blocks, and sizes
> Pass 2: Checking directory structure
> Entry 'kudzu' in /lock/subsys (30001) has deleted/unused inode 30005.
> Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
> Entry 'syslog' in /lock/subsys (30001) has deleted/unused inode 30007.
> Entry 'portmap' in /lock/subsys (30001) has deleted/unused inode 30008.
> Entry 'nfslock' in /lock/subsys (30001) has deleted/unused inode 30009.
> Entry 'syslogd.pid' in /run (38001) has deleted/unused inode 38009.
> Entry 'klogd.pid' in /run (38001) has deleted/unused inode 38010.

I take it that this is after a normal shutdown where you are sure that
the filesystem was unmounted cleanly? It looks like a case where these
files are deleted, but held open by a process.

Could you please try the following:
- "telinit 1" to change into single user mode
- make sure all of the above processes are stopped (check via ps, and
"/etc/rc.d/init.d/foo stop" for each one
- "lsof | grep /var" to see if any files are still open on /var
- umount /var
- e2fsck -f /dev/hdX

Cheers, Andreas
--
Andreas Dilger
http://sourceforge.net/projects/ext2resize/
http://www-mddsp.enel.ucalgary.ca/People/adilger/

2001-11-23 18:50:52

by Marcus Grando

[permalink] [raw]
Subject: Re: Input/output error


Hello,

I have this problem too, i execute fsck more 20 times and not
resolv this problem.

Execute badblocks(8) and not found bad blocks

I note before linux-2.4.15-pre8 begin this problems.

Changelog-pre8:
- Andrew Morton: fix ext3/minix/sysv fsync behaviour.

Maybe? I don?t no.

Regards,

Marcus Grando


On Fri, 23 Nov 2001, Andreas Dilger wrote:

> On Nov 23, 2001 16:43 +0100, Hartmut Holz wrote:
> > On my machine (2.4.15 final) it is the same behaviour. After reboot
> > the lock files (and only the lock files) are corrupt. With 2.4.14 and
> > 2.4.13 everything works fine. gcc 2.96, e2fsck 1.25, aic7896/97
> >
> > e2fsck output:
> > --------------
> > Pass 1: Checking inodes, blocks, and sizes
> > Pass 2: Checking directory structure
> > Entry 'kudzu' in /lock/subsys (30001) has deleted/unused inode 30005.
> > Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
> > Entry 'syslog' in /lock/subsys (30001) has deleted/unused inode 30007.
> > Entry 'portmap' in /lock/subsys (30001) has deleted/unused inode 30008.
> > Entry 'nfslock' in /lock/subsys (30001) has deleted/unused inode 30009.
> > Entry 'syslogd.pid' in /run (38001) has deleted/unused inode 38009.
> > Entry 'klogd.pid' in /run (38001) has deleted/unused inode 38010.
>
> I take it that this is after a normal shutdown where you are sure that
> the filesystem was unmounted cleanly? It looks like a case where these
> files are deleted, but held open by a process.
>
> Could you please try the following:
> - "telinit 1" to change into single user mode
> - make sure all of the above processes are stopped (check via ps, and
> "/etc/rc.d/init.d/foo stop" for each one
> - "lsof | grep /var" to see if any files are still open on /var
> - umount /var
> - e2fsck -f /dev/hdX
>
> Cheers, Andreas
> --
> Andreas Dilger
> http://sourceforge.net/projects/ext2resize/
> http://www-mddsp.enel.ucalgary.ca/People/adilger/
>
> -
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to [email protected]
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/
>

--
Marcus Grando
Administrador de Rede
UNIVALI

2001-11-23 19:10:46

by Tom Eastep

[permalink] [raw]
Subject: Re: Input/output error

On Friday 23 November 2001 07:43 am, Hartmut Holz wrote:
> Rasmus B?g Hansen wrote:
> > On Thu, 22 Nov 2001, Marcus Grando wrote:
> >>On try start syslog deamon occur this errrors "Input/output error" on
> >> many archives /var directory.
> >
> > Try to run fsck on the /var partition. Also you should check the disk
> > for bad blocks. What output do you get from the kernel ('dmesg',
> > /var/log/messages etc.)?
> >
> > It could be a bad disk developing bad sectors.
> >
> > Rasmus
>
> On my machine (2.4.15 final) it is the same behaviour. After reboot
> the lock files (and only the lock files) are corrupt. With 2.4.14 and
> 2.4.13 everything works fine. gcc 2.96, e2fsck 1.25, aic7896/97

A "me too" here (2.4.15 final) -- ran fsck on the /var partition and saw
output similar to that below.
>
> e2fsck output:
> --------------
> Pass 1: Checking inodes, blocks, and sizes
> Pass 2: Checking directory structure
> Entry 'kudzu' in /lock/subsys (30001) has deleted/unused inode 30005.
> Clear? yes
>
> Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
> Clear? yes
>
> Entry 'syslog' in /lock/subsys (30001) has deleted/unused inode 30007.
> Clear? yes
>
> Entry 'portmap' in /lock/subsys (30001) has deleted/unused inode 30008.
> Clear? yes
>
> Entry 'nfslock' in /lock/subsys (30001) has deleted/unused inode 30009.
> Clear? yes
>
> Entry 'random' in /lock/subsys (30001) has deleted/unused inode 30012.
> Clear? yes
>
> Entry 'netfs' in /lock/subsys (30001) has deleted/unused inode 30013.
> Clear? yes
>
> Entry 'autofs' in /lock/subsys (30001) has deleted/unused inode 30014.
> Clear? yes
>
> Entry 'local' in /lock/subsys (30001) has deleted/unused inode 30029.
> Clear? yes
>
> Entry 'syslogd.pid' in /run (38001) has deleted/unused inode 38009.
> Clear? yes
>
> Entry 'klogd.pid' in /run (38001) has deleted/unused inode 38010.
> Clear? yes
>
> Pass 3: Checking directory connectivity
> Pass 4: Checking reference counts
> Pass 5: Checking group summary information
>
> /var: ***** FILE SYSTEM WAS MODIFIED *****
> /var: 23017/66000 files (5.9% non-contiguous), 115364/263168 blocks
>

-Tom
--
Tom Eastep \ [email protected]
AIM: tmeastep \ http://www.shorewall.net
ICQ: #60745924 \_________________________

2001-11-23 20:24:23

by Hartmut Holz

[permalink] [raw]
Subject: Re: Input/output error

Andreas Dilger wrote:


>
> I take it that this is after a normal shutdown where you are sure that
> the filesystem was unmounted cleanly? It looks like a case where these
> files are deleted, but held open by a process.
>


It's a normal shutdown, every filesystem is unmounted cleanly, no
complains from the os.


> Could you please try the following:
> - "telinit 1" to change into single user mode
> - make sure all of the above processes are stopped (check via ps, and
> "/etc/rc.d/init.d/foo stop" for each one
> - "lsof | grep /var" to see if any files are still open on /var
> - umount /var
> - e2fsck -f /dev/hdX


There are a few processes running: news, something called minilgd.Even
with these programs running lsof shows no output. If I kill these
programs and make a normal shutdown, it loocked much better. Only a
complain about the keyboard lock.

If I acted exactly like your advise, there is no problem with the restart.

The processes running in 2.4.15 single user mode are exactly the same as
in 2.4.14.


Regards

Hartmut





2001-11-25 22:50:52

by Horst von Brand

[permalink] [raw]
Subject: Re: Input/output error

Tom Eastep <[email protected]> said:

[...]

> A "me too" here (2.4.15 final) -- ran fsck on the /var partition and saw
> output similar to that below.
> >
> > e2fsck output:
> > --------------
> > Pass 1: Checking inodes, blocks, and sizes
> > Pass 2: Checking directory structure
> > Entry 'kudzu' in /lock/subsys (30001) has deleted/unused inode 30005.
> > Clear? yes
> >
> > Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
> > Clear? yes

[Ad nauseam]

Saw this with 2.4.15pre[89]
--
Horst von Brand [email protected]
Casilla 9G, Vin~a del Mar, Chile +56 32 672616

2001-11-26 00:05:49

by Russell King

[permalink] [raw]
Subject: Re: Input/output error

On Sun, Nov 25, 2001 at 07:50:06PM -0300, Horst von Brand wrote:
> > > Entry 'network' in /lock/subsys (30001) has deleted/unused inode 30006.
> > > Clear? yes
>
> [Ad nauseam]
>
> Saw this with 2.4.15pre[89]

This is the FS bug that Al Viro has fixed for 2.4.15/2.5.0.

--
Russell King ([email protected]) The developer of ARM Linux
http://www.arm.linux.org.uk/personal/aboutme.html