[CentOS] smartmontools

Tue Nov 3 20:21:37 UTC 2009
m.roth at 5-cent.us <m.roth at 5-cent.us>

> At Tue, 3 Nov 2009 11:48:43 -0700 (MST) CentOS mailing list
> <centos at centos.org> wrote:
>> > m.roth at 5-cent.us wrote:
>> >> I'm in the process of rolling out the upgrade from (mostly) 5.3 to
>> >> 5.4. One of my servers started throwing the following:
>> >> Nov  1 05:22:51 <server> kernel:  target4:0:0: FAST-80 WIDE SCSI
>> 160.0
>> >> MB/s DT (12.5 ns, offset 62)
>> >> Nov  1 05:22:51 <server> kernel:  target4:0:1: FAST-80 WIDE SCSI
>> 160.0 MB/s DT (12.5 ns, offset 62)
>> >>
>> >> into my logs every half hour. I don't see anything resembling an
>> error message. The only thing I noted while googling was everyone else
>> spoke
>> >> of "...ns, offset 127", but I have no clue if that's relevant to
>> >> anything. The smartd.conf is the default. I'm not running the debug
>> kernel.
>> >>
>> >> Does anyone have any idea why it's doing this, and, if it's not
>> >> important, how to get it to stop cluttering my logs?
>> >>
>> > What do you see when you run a smartctl -a $DEVICE on the drive that's
>> > choking?
>>
>> Wasn't sure if I should run it on /dev/sdx, or /dev/sdx[#]. I did both,
>> on all three drives, and no errors showing anywhere - the latter two (on
>> /dev/sdb and /dev/sdc) show no uncorrected errors, and no errors
>> corrected by ECC. /dev/sda gives me a lot more output, but says no
errors logged.
>
> smartctl works on the 'bare drive': /dev/sdx.  It makes no sense to run
> it on a partition (and smartctl seems to ignore the partition number).
>
Ah! Thanks.
>>
>> I don't really understand why sda gives so much more info, nor do I
>
> Different make/model/rev/firmware?  smartctl's output depends on what the
> drive is able/willing to tell it in response to various requests.  The
> more 'chatty' (so to speak) the drive, the more output smartctl

Ok, makes sense, though I'd think they were similar. Anyway, thanks for
the info - I've just started using smart tools in the last month or so.
<snip>
          mark