disk error

[Javier Ortiz Guajardo]

i??ve receive this message in syslog
vmunix: LVM: vg[5]: pvnum=16 (dev_t=0x1f366400) is POWERF
AILED
vmunix: LVM: PV 16 has been returned to vg[5].

As you see the disk returned to work.

how can i identify which disk was power failed?.
where i locate the data PV 16 vg[5] and dev_t=0x1f366400

Thanks.

[Ross Zubritski]

Physical Volume 16 in Volume Group 5.

Regards,

RZ

[Pete Randall]

I believe that 0x1f366400 should translate to /dev/dsk/c3t6d0. Start there.


Pete

[Jeff Schussele · Accepted solution]

Hi Javier,

If my hex math is correct that device should be

/dev/dsk/c54t6d4

After the 0x this is how to decode.
First 2 digits are the drive decimal 31 in this case
Next 2 digits are controller in hex - decimal 54 here
Next digit is SCSI ID or "t" value - 6 here
Next didgit is LUN or "d" value - 4 here

So this works out to c54t6d4

HTH,
Jeff

[S.K. Chan]

In addtion to the rest, this may be a low timeout problem with the disk. You can modify it with pvchange (-t option). I think the default is 30sec. You might want to increase it to say 60sec.

[Jeff Schussele]

Oops...should have been...

First 2 digits are the driver - decimal 31 in this case.

Jeff

[Sridhar Bhaskarla]

Jeff is precisely correct. The disk is c54t6d4.

Do a "pvdisplay /dev/dsk/c54t6d4" and get the VG name if your /etc/lvmtab is too large to search.

-Sri

[Steven E. Protter]

Check the power cable on the disk array or disk. Take a look for dangling plugs. We had someone disconnect power to an old HP-6000 disk array while labeling cables for a move.

SEP

[Ross Zubritski]

What the heck is it with people and power cords here lately?? sheesh ;)

Regards,

RZ

[Javier Ortiz Guajardo]

all i want is to translate the message and found which disk was power failed.

I did this:
for message dev_t=0x1f366400
ll /dev/dsk |grep 366400
brw-r----- 1 bin sys 31 0x366400 Mar 2 2001 c54t6d4

for message vg[5]
ll /dev/vgprd4/group
crw-rw-rw- 1 root sys 64 0x050000 Oct 23 2000 /dev/vgprd4/group

for pvnum=16
vgdisplay -v /dev/vgprd4
PV Name /dev/dsk/c39t2d4
PV Name /dev/dsk/c39t2d5
PV Name /dev/dsk/c39t2d6
PV Name /dev/dsk/c39t2d7
PV Name /dev/dsk/c42t10d0
PV Name /dev/dsk/c42t10d1
PV Name /dev/dsk/c42t10d2
PV Name /dev/dsk/c42t10d3
PV Name /dev/dsk/c42t10d4
PV Name /dev/dsk/c42t10d5
PV Name /dev/dsk/c42t10d6
PV Name /dev/dsk/c42t10d7
PV Name /dev/dsk/c54t6d0
PV Name /dev/dsk/c54t6d1
PV Name /dev/dsk/c54t6d2
PV Name /dev/dsk/c54t6d3
PV Name /dev/dsk/c54t6d4

But as you see the c54t6d4 is number 17 if you start to count from 1, If you start from 0 the is 16.

all i need is someone can tell me if this translate is correct to be exactly sure which disk was power failed?

Thanks.

[Rita C Workman]

One last thing to check...got same error...extending the timeout didn't help in my case. Turned out my HBA was flaking in and out and had to be replaced.

Rgrds,
Rita

[Ross Zubritski]

Start from 0.

Regards,

RZ

[Pete Randall]

I need a refresher course in hex! Forget my misguided answer, please.


Pete

[Ross Zubritski]

We still love you Pete! (Sort of)

RZ

[Pete Randall]

Yes, c54t6d4 is correct.

Pete

[Jon Finley]

Assumming that this is your Oracle/Sybase/SQL data disk, AND that the disk is returned to service each time, then it may simply be unable to reply back within the default 30 seconds.

Rita's correct though. If changing the timeout value (pvchange) doesn't resolve the problem, you may be looking at replacing the disk.

And... yes you've found the correct disk at c56t6d4.

Jon