3ware raid

Carroll Kong me at carrollkong.com
Tue Oct 26 15:40:34 PDT 2004


Well, in the normal cases (where the disk is legimately dying) you can
usually replace it and be okay.  (recalling one instance from the massive
thread regarding a 3ware 7506 with maxtor disks that had issues but once
they replaced bad disks they were a-ok).

You can use smartmontools to help you run tests on the hdds under a 3Ware
controller while still in FreeBSD.  It works in 4.X (with the latest
snapshot of smartmontools).

In the other cases where the disk appears fine, but still randomly pulls
disks into DEGRADED mode, you have to find out why.

In some cases it's because the power drop is too significant, which means
you need to swap out your power supply.  In other cases, the IDE cabling is
legitimately bad (in my case, I highly suspect it is the IDE backplane,
which is part of the disk-to-controller data path so I put bad cabling in
quotes there).  In short, 3Ware makes a note that you must ensure the entire
data path is squeaky super clean.  Of course, the problem could be the
controller itself...

3Ware claims it takes a lot of punishment to DEGRADE the disks.  There are
some procedures to pull out the extended diagnostics from 3Ware controllers
via the 3ware command line utility and if you send them to 3Ware they can
try to tell you more information.

In my case, 3Ware copped out and gave me some BS excuse for why my system
had issues.



- Carroll Kong
----- Original Message -----
From: "Marc G. Fournier" <scrappy at hub.org>
To: "Carroll Kong" <me at carrollkong.com>
Cc: <freebsd-stable at freebsd.org>
Sent: Tuesday, October 26, 2004 6:27 PM
Subject: Re: 3ware raid


> On Tue, 26 Oct 2004, Carroll Kong wrote:
>
> > 4)  Monitor the system carefully.  Apparently in DEGRADED mode, the
> > system has a high chance of complete lockups if left unattended.  It is
> > important to note that the 3Ware controller is very sensitive.  People
> > have gone into DEGRADED mode randomly because of slight power drops or
> > "bad" IDE cabling.
>
> How do you test (and fix) something like this?  revive the drive and hope
> the same drive doesn't happen again?
>
>
> ----
> Marc G. Fournier           Hub.Org Networking Services
(http://www.hub.org)
> Email: scrappy at hub.org           Yahoo!: yscrappy              ICQ:
7615664
>



More information about the freebsd-stable mailing list