Re: unexplained SATA errors (RAID-5, LVM in use)
From: Steve Wolfe (unt_at_codon.com)
Date: 03/03/05
- Next message: Folkert Rienstra: "Re: How I built a 2.8TB RAID storage array"
- Previous message: Jules: "Re: skipping tape errors with tar?"
- In reply to: Dan Christensen: "unexplained SATA errors (RAID-5, LVM in use)"
- Next in thread: Dan Christensen: "Re: unexplained SATA errors (RAID-5, LVM in use)"
- Reply: Dan Christensen: "Re: unexplained SATA errors (RAID-5, LVM in use)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Wed, 2 Mar 2005 18:01:05 -0700
> I'm running linux kernel 2.6.10 on a system with three 160G Samsung
> Spinpoint SATA drives, sda, sdb and sdc, each partitioned into four or
> five partitions.
>
> I got the following error today:
>
> Mar 1 04:41:10 localhost kernel: ata2: status=0x51 { DriveReady
SeekComplete Error }
> Mar 1 04:41:16 localhost kernel: ata2: error=0x40 { UncorrectableError }
>
> No other information was logged, and the same error appeared three
> more times within 7 seconds.
>
> I don't know for sure which drive is ata2, so I ran
>
> badblocks -s -o /tmp/badblocks.sd$i$j /dev/sd$i$j
>
> for i=a,b,c and for all appropriate j, but all of the output files
> were empty. During the badblocks run, the same error appeared once
> more.
Look at the output of smartctl -a /dev/hd{x}. Badblocks won't necessarily
find something. Here's why:
Every drive has a reserve of unallocted sectors, often around 200. When
the drive detects that a spot on the disk has failed, it will silently remap
that sector to one of the reserved sectors. When it does that, the data in
the sector may or may not have been lost. But, the next time you scan for
errors, the drive uses the good (reserved) sector instead of the bad one,
and you don't know about it. However, as long as SMART has been enabled on
the disk, the event will be logged, and you can find out about it. If the
drive has remapped ANY sectors, it's prudent to get a new drive. The exact
wording depends on the drive and manufacturer, but look for items like
"reallocated sectors", "offline uncorrectable", "reallocated event count",
etc..
Now that I've said that, the error could also be in the electronics of the
drive, in which case you're kind of up the creek without a paddle.
steve
- Next message: Folkert Rienstra: "Re: How I built a 2.8TB RAID storage array"
- Previous message: Jules: "Re: skipping tape errors with tar?"
- In reply to: Dan Christensen: "unexplained SATA errors (RAID-5, LVM in use)"
- Next in thread: Dan Christensen: "Re: unexplained SATA errors (RAID-5, LVM in use)"
- Reply: Dan Christensen: "Re: unexplained SATA errors (RAID-5, LVM in use)"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|