vxfs: mesg 056: vx_dataioerr - /dev/vgarr_01/lvarr_01 file system file data read error

[Serhiy Cherpatyuk]

Hello,

Oracle can't read the file from the FC10's disk.

In the syslog we have next message:

vxfs: mesg 056: vx_dataioerr - /dev/vgarr_01/lvarr_01 file system file data read error

I try to check the suspicious disk.
The result of the Info tool (STM):

-- Information Tool Log for SCSI Disk on path 8/8.8.0.255.0.12.0 --
Log creation time: Tue Aug 26 16:53:39 2003
Hardware path: 8/8.8.0.255.0.12.0
Product Id: ST336704FC Vendor: SEAGATE
Device Type: SCSI Disk Firmware Rev: HP04
Device Qualifier: SEAGATEST336704FC Logical Unit: 0
Serial Number: 3CD0ZH71000021202597
Capacity (M Byte): 34732.89
Block Size: 512
Max Block Address: 71132959
Error Logs
Total Retries: 0 Buffer Overruns: N/A
Read Reverse Errors: N/A Buffer Underruns: N/A
Write Errors: 0 Non-Medium Errors: 55
Verify Errors: 0

The result of the Verify Tool - Successful
The result of the Exersize Tool (maximum stress level, non-destructive) - Successful

Seems that the HDD is OK (but I don't try to run "fsck -F vxfs -y -o full" yet).

How can I solve this problem?

Thank you.

PS the output of the HealthCheck attached.

[Eugeny Brychkov]

Serhiy,
ITRC does not handle .gz files. Please rezip with windows winzip.
To check disk readability try with dd. If read will pass ok (without 'I/O error') then most probably there're problems with VxFS maps. I think you can try running fsck without keys (without fixing) to see if it will find anything
Eugeny

[Serhiy Cherpatyuk]

Eugeny,

thank you for answer. Healthcheck script runs the "dd" (instead of me) but with bs=512. The result is:

/dev/rdsk/c4t12d0
1+0 records in
1+0 records out
0000000 4445 4645 4354 3031 b1e2 ab10 8000 0000
0000010 b1e2 ab18 8000 0000 0000 0000 0000 0000
0000020 0000 0000 0000 0000 0000 0000 0000 0000
*
0000200
1+0 records in
1+0 records out
0000000 4445 4645 4354 3031 b1e2 ab10 8000 0000
0000010 b1e2 ab18 8000 0000 0000 0000 0000 0000
0000020 0000 0000 0000 0000 0000 0000 0000 0000
*
0000200

seems there is no bad blocks. But in in any case I'll try to run dd with bs=64.

Serhiy.

[Serhiy Cherpatyuk]

Hello again,

what is the plausible reason of this type of data corruption?

Serhiy.

[Eugeny Brychkov]

It's hard to say now what's the reason. Anyway non-medium errors counted (55) are suspicious. Looking to your dd output I guess it reads only first sector of the drive (1+0 means that it read 1 record of 512 bytes). I suggest to check whole disk with block size bs=1024k and watch it to return something like
138931+138931
and not 1+0.
Do you have command view for disk systems installed? You can try its diagnostics commands in addition to STM's
Eugeny