Re: problems with disks

From: Michael Vilain (vilain_at_spamcop.net)
Date: 12/07/03 

Date: Sun, 07 Dec 2003 10:32:46 -0800

In article <bquva2$sgu$02$1@news.t-online.com>,
 Julian Thomas <julian.thomas@t-online.de> wrote:

> Hi!
>
> i've got a D1000 Diskarray with 12 disks (10x 9.1GB, 2x 18GB)
> Last night i ran the verify from the 'format' command an got many errors.
>
> # format
>
> [...]
>
> Specify disk (enter its number): 11
> selecting c1t12d0
> [disk formatted]
>
> [...]
>
> format> analyze
>
> [...]
>
> analyze> verify
> Ready to verify (will corrupt data). This takes a long time,
> but is interruptable with CTRL-C. Continue? y
>
> pass 0
> Dec 6 20:56:17 U10-S1 scsi: WARNING: /pci@1f,0/pci@1/scsi@1 (glm0):
> Dec 6 20:56:17 U10-S1 Target 12 reducing sync. transfer rate
> 4923/26/7
>
> pass 1
> Data miscompare error (expecting 0xfff0, got 0x91e82000) at 18/6/84,
> offset = 0x0.
> 4923/26/7
>
> Total of 0 defective blocks repaired.
> analyze>
>
> [...]
>
> can i still use thus disk without risk? or should i replace this disk?
> (both errors came on several disks)
>
> The /var/adm/messages contains lot of these entries:
>
> Dec 6 20:44:59 U10-S1 scsi: [ID 107833 kern.warning] WARNING:
> /pci@1f,0/pci@1/scsi@1 (glm0):
> Dec 6 20:44:59 U10-S1 Target 4 reducing sync. transfer rate
> Dec 6 20:44:59 U10-S1 glm: [ID 923092 kern.warning] WARNING:
> ID[SUNWpd.glm.sync_wide_backoff.6014]
> Dec 6 20:45:02 U10-S1 scsi: [ID 107833 kern.warning] WARNING:
> /pci@1f,0/pci@1/scsi@1/sd@0,0 (sd15):
> Dec 6 20:45:02 U10-S1 corrupt label - wrong magic number
> Dec 6 20:45:02 U10-S1 scsi: [ID 107833 kern.warning] WARNING:
> /pci@1f,0/pci@1/scsi@1/sd@1,0 (sd16):
> Dec 6 20:45:02 U10-S1 corrupt label - wrong magic number
>
> what can i do to resolv this problem?
>
> thanks in advance
>
> Julian
>

Why would you want to use the disk at all if it got fatal write errors?
Is your data that worthless?

The error you posted isn't a disk failure but the hardware telling you
that there's a slow device in the SCSI chain and that it's 'throttling'
it's transfer rate so as not to overwhelm the slow guy. If you're
mixing SCSI (III and IV) in a chain, this can happen.

Put each type of SCSI on their own controller. 

-- 
DeeDee, don't press that button!  DeeDee!  NO!  Dee...