Disks powerfailed?

[Brad Andriese]

System: G40 HP-UX 10.10

The messages below used to appear in the syslog once a week or so. All of the sudden, they're happening every minute or so. The messages seen below repeat in the same pattern with different times. What causes this? If the drive was bad, why does the system think everything is back to normal a split-second or the same second later?

May 14 16:20:54 ginger vmunix: Disk at 52.3.0 is not responding. Check device, power, and cables.
May 14 16:20:54 ginger vmunix: LVM: vg[1]: pvnum=0 (dev_t=0x1c013000) is POWERFAILED
May 14 16:20:54 ginger vmunix: LVM: PV 0 has been returned to vg[1].

[A. Clay Stephenson · Accepted solution]

Hi Brad,

From the looks of things you have a drive in immenent danger of complete failure. I hope it's mirrored so that replacement will be a breeze. As to why is comes back after a little while: Loose power connection? Loose/poor scsi connection? Excessive noise on the buss - possibly bad terminator? I think however that the drive is on its last legs. If you're under maintenance, I'd get Mr. Goodwrench on site rightaway; if not, I'd start looking over the HP used equipment market.

Regards, Clay.

P.S. There is a really good chance that this puppy won't survive a cold boot - so if you are not mirrored - backup (and probably if you are).

[Printaporn_1]

Hi,

I user to found this in 10.20 w/o disk failure
the cause can be.

SCSI timeout , you can

1. install latested SCSI , file system patch
2. check terminator
3. increase timeout.

[Vincenzo Restuccia]

Is the SCSI termination.See http://forums.itrc.hp.com/cm/QuestionAnswer/1,1150,0x43fea12d6d27d5118fef0090279cd0f9,00.html

[Brad Andriese]

Thanks for the replies. There are 5 disks internaly. Since these are internal disks on the G40, I don't think this is a termination problem. The disks are mirrored. This disk may be failing, but that doeesn't explain why I would still get these messages, over a period of several months.

Changing the time-out seems reasonable, but it makes me wonder why it is not correct to begin with?

[Jim Moffitt_1]

I've had problems like that in the past on a K580 Model30 disk array and it turned out to be a controller. My advice is to place a hardware call.

[T S_1]

I agree with Jim. I had a similar issue in the past, I replaced the disk, then the controller and after that changed the timeout on the disk.
Did not get any errors after that.

E. F.