LinuxQuestions.org

LinuxQuestions.org (/questions/)
-   Linux - Software (https://www.linuxquestions.org/questions/linux-software-2/)
-   -   kernel I/O error (https://www.linuxquestions.org/questions/linux-software-2/kernel-i-o-error-465599/)

UltraSoul 07-19-2006 12:38 PM

kernel I/O error
 
Hi, all

I got the following /var/messages, I wonder if it is a harmless messages. If it is harm to my system, how to troubleshoot the problem. The system info is as follows:




Jul 16 04:03:27 sys77 kernel: I/O error: dev 08:21, sector 517
Jul 16 04:03:28 sys77 kernel: Directory sread (sector 0x205) failed


I found the error messages always occur around 04:03 every day.

raskin 07-19-2006 01:16 PM

Seems like your SCSI disk (sdb5 partition, maybe; 'ls -l /dev/sd* | grep 21' will help) is feeling bad. I'm afraid you have lost one of directories. Do you have backups of all the rest? After you do, try investigating fs integrity on disks with fsck, and maybe physical integrity of disk surface.

UltraSoul 07-22-2006 07:21 AM

I found an interesting thing.
 
Hi, raskin

The I/O errors seem to occur at 04:03 every day. I think if it is related to
cron job of every day.

UltraSoul 07-22-2006 07:35 AM

I found an interesting thing.
 
Hi, raskin

The I/O errors seem to occur at 04:03 every day. I think if it is related to
cron job of every day.

raskin 07-22-2006 03:11 PM

That just means that some of your cron jobs references damaged directory. So you can even find it. Have you already backed up the data? Have you already checked the disk?

UltraSoul 07-25-2006 03:01 AM

#ls -l /dev/sd* | grep 21
brw-rw---- 1 root disk 8, 21 Jun 25 2004 /dev/sdb5

raskin 07-25-2006 04:08 AM

Wow, I guessed. So, run 'cat /proc/mounts | grep sdb', backup data, then umount the partition on sdb5 and run fsck on it.

UltraSoul 07-25-2006 08:34 AM

Hi, raskin

I wonder what meaning about 21?
Run #ls -l /dev/sd* | grep 21
brw-rw---- 1 root disk 8, 21 Jun 25 2004 /dev/sdb5

The output above means the disk I/O error occured at root disk 8,21 with /dev/sdb5 mounted, is it right?

And cat /proc/mounts as follows:
#cat /proc/mounts
rootfs / rootfs rw 0 0
/dev/root / ext3 rw 0 0
/proc /proc proc rw 0 0
none /dev/pts devpts rw 0 0
usbdevfs /proc/bus/usb usbdevfs rw 0 0
/dev/sda1 /boot ext3 rw 0 0
none /dev/shm tmpfs rw 0 0
automount(pid3846) /d_map autofs rw 0 0
/
I can not grep sdb.

From the result above, I can not monitor the /dev/sda5 is mounted.

raskin 07-25-2006 08:41 AM

Well, are you investigating from sys52?

21.. It is minor number. Every block device has major and minor number, which are like ip address, and /dev/sdb5 is like domain name. And then /dev/sdb5 can be mounted to become some directory.

UltraSoul 07-25-2006 08:53 AM

I have ran sysreport from sys50.
And How to decide the filesystem mounted on /dev/sda5.


Please teach me. Thank you very much.

UltraSoul 07-25-2006 10:52 AM

# more /proc/partitions
major minor #blocks name

8 0 8388608 sda
8 1 7711168 sda1
8 2 674730 sda2

I can not see the minor 21 0f sda5. i wonder if it means the /dev/sda5 has some problem and causes the system can not locate
the /dev/sda5

raskin 07-25-2006 01:49 PM

You need sdb5. And you need to investigate what is /dev/sdb5 on sys52 - look at you log..

UltraSoul 07-25-2006 07:53 PM

# fdisk -l
<---- Disk: /dev/sda ---->

Disk /dev/sda: 73.4 GB, 73407865856 bytes
255 heads, 63 sectors/track, 8924 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1058 8393962+ 82 Linux swap
/dev/sda3 1059 8924 63183645 83 Linux

<---- END ---->
<---- Disk: /dev/sdb ---->


<---- END ---->
<---- Disk: /dev/sdc ---->


<---- END ---->



I found the /dev/sdb and /dev/sdc have not been recognized the host.

Matir 07-25-2006 08:18 PM

After a fresh boot, what does 'dmesg | grep sd' show?

UltraSoul 07-25-2006 11:14 PM

I wonder if the /dev/sdb should be replaced.

Or I can run "#fsck -y /dev/sdb5" for trying to fix the I/O error.

Because the /dev/sdb5 is unmounted. I wonder if i can run the command when system running. Or it is the best way to go to single user mode to run the fsck.

Matir 07-25-2006 11:39 PM

If fdisk -l sees no /dev/sdb5, I'm wondering what kind of problem you are suffering. What is the contents of /proc/partitions?

raskin 07-26-2006 04:01 AM

Did you run fdisk -l on sys52 ???

UltraSoul 07-26-2006 09:12 AM

#fdisk -l
<---- Disk: /dev/sda ---->


Disk /dev/sda: 73.4 GB, 73407865856 bytes
255 heads, 63 sectors/track, 8924 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes

Device Boot Start End Blocks Id System
/dev/sda1 * 1 13 104391 83 Linux
/dev/sda2 14 1058 8393962+ 82 Linux swap
/dev/sda3 1059 8924 63183645 83 Linux

<---- END ---->
<---- Disk: /dev/sdb ---->


<---- END ---->
<---- Disk: /dev/sdc ---->


<---- END ---->

UltraSoul 07-26-2006 09:13 AM

#more proc/partitions
major minor #blocks name rio rmerge rsect ruse wio wmerge wsect wuse runni
ng use aveq

8 0 71687369 sda 480938 1011060 11824556 943580 681049 4920186 44845318
1775628 0 4387990 3028338
8 1 104391 sda1 321 17329 35300 1170 71 60 262 62160 0 51210 63330
8 2 8393962 sda2 27 66 336 70 109 12233 98736 284990 0 78930 285060
8 3 63183645 sda3 480395 993005 11787210 940970 680869 4907893 44746320
1428488 0 4308930 2678548
8 16 1048575 sdb 32 96 256 30 0 0 0 0 0 30 30
8 32 1048575 sdc 105683 19585 125440 26770 6 0 6 220 0 16410 26990

UltraSoul 07-26-2006 09:16 AM

My Linux Box suffers the following I/O error. From the analysis above, I think the I/O error occurs at the 5th slice of the secondary scsi hard disk. I want to know how to fix the I/O error
--- fsck ?

Jul 18 04:03:20 sys52 kernel: Directory sread (sector 0x1f5) failed
Jul 18 04:03:20 sys52 kernel: I/O error: dev 08:21, sector 501

Matir 07-26-2006 09:32 AM

According to your system, /dev/sdb has no partitions, so the error with 08:21 is very odd. What is on the sdb disk?

UltraSoul 07-26-2006 10:04 AM

Some Read failure to sdb.


I think the sdb can not be recognized by kernel.

raskin 07-26-2006 10:17 AM

well, maybe post dmesg from sys52..

UltraSoul 07-27-2006 07:44 AM

I found sys52 just one SCSI disk connected. I stopped the cron job, then the dev I/O error disappeared.

Thank you.


All times are GMT -5. The time now is 07:48 PM.