From mboxrd@z Thu Jan  1 00:00:00 1970
From: Chris Murphy <lists@colorremedies.com>
Subject: Re: Inactive arrays
Date: Tue, 13 Sep 2016 13:52:12 -0600
Message-ID: <CAJCQCtS-aYNgy7MM990CTKFqcNk8bf+ggzX3QRrD+yQZfdVxnA@mail.gmail.com>
References: <CAHscji0ddZT-aq5fEp8aG030+F7UqGzMgRnm4QP1Lr1AsGy61Q@mail.gmail.com>
 <57A07345.4040708@youngman.org.uk> <CAHscji0BQuyKQbLYZ9Ah16hHday_seTaajm8LOn6+HRFqinTyQ@mail.gmail.com>
 <CAHscji3C4Cygs=xh4d4PtREp9mGSBjNS0o7SatW_QotzYShA_Q@mail.gmail.com>
 <CAHscji3HdHaJAvwtYNU=Ykc_qohBkfFfrbP0M=FNhuFMK+d-Jg@mail.gmail.com>
 <CAHscji35kySP4Q8cUpYCXR2P9DBwPcYeWjr_HX==TNX9CPW9NA@mail.gmail.com>
 <CAHscji2k9FyrOs+96XQWEUWU3W0EWxkanaHakjsGhij0hXiVGw@mail.gmail.com>
 <57D72092.20704@youngman.org.uk> <CAHscji3KfFfO8CxZTbLX6u-bfA57pOR5i6qvc0NhNODH5tGeEg@mail.gmail.com>
 <CAJCQCtTzPpt0krPxCZY_TB-Xa6PcG0YsdDPnneq5dEB30wy8hw@mail.gmail.com> <CAHscji0H7kEDzhg3KpUYrdk5MA06k-F5Xe95=EL7mo-_mS3y5A@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Return-path: <linux-raid-owner@vger.kernel.org>
In-Reply-To: <CAHscji0H7kEDzhg3KpUYrdk5MA06k-F5Xe95=EL7mo-_mS3y5A@mail.gmail.com>
Sender: linux-raid-owner@vger.kernel.org
To: Daniel Sanabria <sanabria.d@gmail.com>
Cc: Chris Murphy <lists@colorremedies.com>, Wols Lists <antlists@youngman.org.uk>, Linux-RAID <linux-raid@vger.kernel.org>
List-Id: linux-raid.ids

On Tue, Sep 13, 2016 at 1:43 PM, Daniel Sanabria <sanabria.d@gmail.com> wrote:
>> This is a problem. What do you get for
>>
>> cat /sys/block/sdc/device/timeout
>
> [root@lamachine ~]# cat /sys/block/sdc/device/timeout
> 30
> [root@lamachine ~]# cat /sys/block/sdd/device/timeout
> 30
> [root@lamachine ~]# cat /sys/block/sde/device/timeout
> 30
> [root@lamachine ~]#

Common and often fatal misconfiguration. Since the drives don't
support SCT ERC, the command timer needs to be changed to something
higher. Without the benefit of historical kernel messages, it's
unclear if there have been any link resets that'd indicate improper
correction for bad sectors on the drives.




>
>> Anyone specifically familiar with WDC Greens, and if the lack of SCT
>> ERC can be worked around in the usual way by increasing the SCSI
>> command timer value? Or is there also something else? I vaguely recall
>> something about drive spin down that can also cause issues, does that
>> need mitigation? If no one chimes in, this information is in the
>> archives, just search for 'WDC green' and you'll get an shittonne of
>> results.
>
> In another thread I found Phil Turmel recommending to change the
> timeout value like this:
>
> for x in /sys/block/*/device/timeout ; do echo 180 > $x ; done
>
> Is that what you guys are talking about when mentioning the SCT/ERC issues?

Yes. You should do that.




>
>> OK so the next thing I want to see is why you're getting these
>> messages from parted when you check sdc and sde for partition maps. At
>> the time you do this, what do you see in kernel messages? Maybe best
>> to just stick the entire dmesg for the current boot up somewhere like
>> fpaste.org or equivalent.
>
> https://paste.fedoraproject.org/427719/37952531/

Yeah that looks like a recent boot; if that's a boot where you'd run
parted and got those errors on read, then I don't have a good
explanation why you're getting parted errors that don't have matching
kernel messages, i.e. something from libata about the drive not liking
the command or not properly reading from the drive, etc.

What do you get for gdisk -l <dev> for each of these drives?



-- 
Chris Murphy