I/O Error

[Eddie Salvatierra]

The following error message has appeared in my Syslog:

Aug 6 12:48:54 hn227 vmunix: Page I/O error(1) while swapping from disk for device 0x03000000,
Aug 6 12:48:54 hn227 vmunix: block #508218, page addr = 2f1.401ef000mna
Aug 6 12:48:54 hn227 vmunix: Making the above swap block unavailable for future use. Please check
Aug 6 12:48:54 hn227 vmunix: for any possible hardware/firware errors on the swap
Aug 6 12:48:54 hn227 vmunix: device IMMEDIATELY

The Server is a 9000/813/D330 running HPUX 10.20

I have checked /dev directory and I cannot map the major and minor number to a disk.

swapinfo output is below:

Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 262144 52384 209760 20% 0 - 1 /dev/vg00/lvol2
localfs 262144 0 262144 0% 262144 0 1 /db/paging
reserve - 50656 -50656

Has anybody encountered this error ??? Why is it that I cannot map the major and minor numbers ??

Cheers
Eddie Salvatierra

[linuxfan]

Hi Eddie,

Found a HP doc S1100002373 which resembles your problem. Hope this is helpful

/Begin/

The /var/adm/syslog/OLDsyslog.log file contains many of the following
messages:

Page I/O error(1) while swapping from disk for device 0x03000000,
block #4210812, page addr = 3077.7b03b000
Making the above swap block unavailable for future use. Please
check for any possible hardware/firware errors on the swap device
IMMEDIATELY

These messages are followed by this system panic:

vfdswapi: syncpageio detected an error

Since I am running multiple swap devices, how can I determine which
device has the hardware problem?


CONFIGURATION
Operating System - HP-UX
Subsystem - System Interrupts

RESOLUTION
From a software perspective, the block number in the error message
(4210812) will help you to identify the problem hardware device.

Perform the following steps to locate the problem device:

1. Omit the last 14 bits of 4210812 to get the swap chunk number:

4210812 / 16384 = 257

Note: Use the swapconf.pl script (which you should have in the
q4lib) to determine on which disk this swap chunk resides.

2. Run q4 on the crashdump.

3. Type 'include swapconf.pl' at the q4 prompt.

4. Type 'run SwapConf > /tmp/swapconf' at the q4 prompt.

5. Type 'exit' at the q4 prompt.

6. Search for the 'swaptab' table in the /tmp/swapconf file.

7. Use the swap chunk number as index and determine the value of
dev_t for this swap chunk:

index 257 -> dev_t=0x40010001

Note: 'dev_t' contains a major and a minor number:

major: first 2 hex digits: ponits to driver -> device type

minor: other hex digits: first 2 -> vgnumber -,
last 4 -> lvolnumber -> Ox40 -> 64 decimal ->
the device is a lvol (0x1f for example, would be a disk)
-> 010001 -> vg01/lvol1

8. Execute the following command to identify the disk:

# lvdisplay -v /dev/vg01/lvol1 | more

Note: In this case, c0t5d0 is the problem disk.

You can do one of two things:

A. Log a hardware call to check/replace the disk.

--OR--

B. Perform the hardware check.


If you decide to perform a hardware check:

1. Determine the swap devices:

# swapinfo -t

Note: The output in this case reports
/dev/vg00/lvol2 and /dev/vg01/lvol1.

Kb Kb Kb PCT START/ Kb
TYPE AVAIL USED FREE USED LIMIT RESERVE PRI NAME
dev 524288 7260 517028 137; 0 - 1 /dev/vg00/lvol2
dev 524288 7080 517208 137; 0 - 1 /dev/vg01/lvol1
reserve - 139948 -139948
total 1048576 154288 894288 1537; - 0 -

2. Execute the following comamnds to identify the disk:

A. # lvdisplay -v /dev/vg00/lvol2 | more

B. # lvdisplay -v /dev/vg01/lvol1 | more

Note: In this example, /dev/vg00/lvol2 corresponds to disk c0t6d0,
and /dev/vg01/lvol1 corresponds to disk c0t5d0.

Finally, via online (cstm/mstm/xstm) or offline diagnostics (boot from
support CD-ROM, run ODE) review the disk logs. You should see the same
problems that you saw after completing Steps 1 through 8 (of the first
procedure).

Once again, you can log a hardware call to check/replace the disk.

/End/

This should help you in determining the faulty disk

-HTH
Ramesh