HP-UX Error: 5: I/O error

[Francis Adekunle]

September 11, 2003 19:43 PM GMT

--------------------------------------------------------------------------------
I first noticed this when a user was unable to load data into an oracle table. Below are the results of some command I 'm using to check the status of the disk......Do I have a bad disk...? help pls...
HP-UX 10.20 D320

#bdf
bdf: /u04: I/O error

#ioscan -fnC disk

disk 7 8/4.3.0 sdisk CLAIMED DEVICE SEAGATE ST318275LC
/dev/dsk/c0t3d0 /dev/rdsk/c0t3d0



#vgdisplay -v /dev/dsk/vg05
vgdisplay: Cannot open the control file "/dev/dsk/vg05/group":
No such file or directory
vgdisplay: Cannot display volume group "/dev/dsk/vg05".

#pvdisplay -v /dev/dsk/c0t3d0 | more

PV Status = unavailable
--- Distribution of physical volume ---
LV Name LE of LV PE for LV
/dev/vg05/u04 4340 4340


#vgdisplay -v | more

VG Name /dev/vg05
VG Write Access read/write
VG Status available
Max LV 255
"
"
--- Logical volumes ---
LV Name /dev/vg05/u04
LV Status available/syncd
LV Size (Mbytes) 17360
Current LE 4340
Allocated PE 4340
Used PV 1


--- Physical volumes ---
PV Name /dev/dsk/c0t3d0
PV Status unavailable
Total PE 4340
Free PE 0
Autoswitch On


My physical disk is c0t3d0...

[Jeff Schussele]

Hi Francis,

Possible that you have a bad disk, but more importantly the file containing the VG info - /dev/dsk/vg05/group is missing.
That would prevent the system from accessing vg05. If missing you need to restore it from backup & investigate why it went missing.

Rgds,
Jeff

[Steven E. Protter]

You have a disk on the blink.

dmesg

If you have /etc/lvmtab, the utility I'm attching might help.

SEP

[Helen French]

You may have to look from both hardware and software side:

1) To make sure your disk is okay, use "stm". Run information and exercise tools and see if the disk reports any error. Also run dd commad:

# dd if=/dev/dsk/c0t3d0 of=/dev/null bs=1024K

2) Since vgdisplay reported 'group' file missing, this can be a software (LVM) error too. Check if you have the file exists and the permissions. You can restore or recreate this file.

[Stuart Abramson_2]

your vgdisplay command is bad:

. NOT:

. vgdisplay -v /dev/dsk/vg05

. SHOULD BE:

. vgdisplay -v /dev/vg05

However, you do have a bad disk, or cable or something, because of the messages:

. "PV Status unavailable"

I've never seen that before.

Check syslog.log; check dmesg; do a diskinfo:

. diskinfo /dev/rdsk/c0t3d0.

Go LOOK at the disk and check for lights.

[Jeff Schussele]

Well, I'll be...

I missed that too. I'm so used to using *just* the VG name - i.e.

vgdisplay vg05

That I didn't even notice the erroneous path.

So follow the latest advice & do a diskinfo first & IF it reports no or bad info, then disk is bad. If correct info do a dd on it to test read it & also if it's bad to light it's activity light to spot it - like:

dd -if=/dev/dsk/c0t3d0 -of=/dev/bull bs=2k

Rgds,
Jeff

[Jeff Schussele]

Ha..ha..Freudian slip maybe?

That would be /dev/null...not bull

But who knows - maybe I'm critiquing myself 8~P

Cheers,
Jeff (If you can't laugh at yourself) Schussele

[Francis Adekunle]

Thanks guys....
The problem is two folds... I do not want to replace the disk if it's not bad...Also, I do need to know which physical disk to replace.So , I need to be able to map the c0t3d0 to a physical disk in the external Jamaica box....

Below are the results of some of the y 'all suggestions...

#dmesg | more
vxfs: mesg 055: vx_metaioerr - /dev/vg05/u04 file system meta data read error
vxfs: mesg 016: vx_ilisterr - /u04 file system error reading inode 3683
"
"
vxfs: mesg 017: vx_ilisterr - /u04 file system inode 2 marked bad
vxfs: mesg 015: vx_ilisterr - /u04 file system can't mark inode 2 bad
vxfs: mesg 031: vx_disable - /dev/vg05/u04 file system disabled

Jeff said I need to restore from data....but there is no mount point available to restore to ...it's missing.

# /usr/sbin/diskinfo -v /dev/rdsk/c0t3d0
SCSI describe of /dev/rdsk/c0t3d0:
vendor: SEAGATE
product id: ST318275LC
type: direct access
size: 17783240 Kbytes
bytes per sector: 512
rev level: HP04
blocks per disk: 35566480
ISO version: 0
ECMA version: 0
ANSI version: 2
removable media: no
response format: 2
(Additional inquiry bytes: (32)41 (33)4b




# dd if=/dev/dsk/c0t3d0 of=/dev/null bs=1024K

dd: invalid number specification in option

...Even though I 've a backup of this system, I need to be sure if this is a hardware or Software failure..

Thanks for all the reponse..

[Kent Ostby]

Francis --

I've been doing support on HP boxes for about 10 years now and I've seen ERRNO 5 on numerous occassions and its always been HW as far as I can recall.

As far as your "dd" command, I dont think it likes your blocksize. Try:

# dd if=/dev/dsk/c0t3d0 of=/dev/null bs=64K

Best regards,

Kent M. Ostby

[Francis Adekunle]

It seems the dd command doesn't like me.....even after changing the block size, it still not working...

However, I was able to determine the physical disk affected. I ran the diskinfo command on the console and wathed the light...

I 'm leaning towards disk replacement..

This is the expected process for the disk replacement:
1. Replace the failed disk.
2. Boot the system.
3. Restore the LVM header on the disk:
# vgcfgrestore -n /dev/vg05 /dev/rdsk/c1t3d0
4. Activate the volume group:
# vgchange -a y /dev/vg05
5. Check, which logical volumes have extents on the hard disk:
# pvdisplay -v /dev/dsk/c1t3d0
6. The backup of these logical volumes has to be restored. For logical volumes with file systems a newfs has to be executed prior to restoring the data.


Guys, any alternate suggestions ...?

Appreciate all the contributions.....it's highly educative and informational....