Re: [gentoo-user] smartctrl drive error @60%

[David Haller <gentoo@dhaller.de>]

Hello,

On Wed, 25 Jun 2014, Dale wrote:
>David Haller wrote:
>> On Wed, 25 Jun 2014, Dale wrote:
>> Yeah. Oh, and I had a "clean" smart until a few days ago, luckily I
>> alread had a WD Red (WD40EFRX) drive waiting when this attrib jumped
>> from 0 to: 5 Reallocated_Sector_Ct 0x0033 087 087 036 Pre-fail Always
>> 17688 Other Seagates (a few 1.5T drives) have also made me trouble,
>> the 2T Samsung already relabeled and sold as a Seagate but with
>> Samsung in the FW though is still ok. [..] 
>
>I was wondering about how that would be updated since a lot of that
>stuff requires windoze. 

I had the "fun" with a couple of those 2TB Samsung drives a while ago
(Jan 2012?). Samsung had some .iso files available, which you could
write to a CD/DVD/USB-Shtik, and then boot from them. For me, the
biggest problem was _which_ of the 4 HD204UI and 2 HD203WI I had and
have needed the update... I don't remember the SW displaying serial
numbers... *gah* Anyway, I've got it sorted out and updated those that
needed the update. With Seagate/WD/HGST/Toshiba I've no experience. 
And I still am grumpy about Samsung selling off their HDD stuff to
Seagate.

BTW: in german, we have a saying for Seagate drives: "sie geht oder
sie geht nicht", basically spoken as "sea gate odr sea gate nicht"
meaning "she works or she won't"... But, talk about IBM "deathstars",
and WD has also a record, basically, all HDDs are much alike nowadays,
and have been for years. There's always a bad batch somewhere ...

Seagate: Seagate drives (???), ex-Maxtor drives, ex-Samsung drives
WD: bought Hitachi GST (formerly IBM), i.e. WD + ex-IBM/HGST drives
Toshiba: new player, no warranty for bulk drives, unknown for the desktop

I miss the days when you still had a real choice (WD, Seagate, Maxtor,
Samsung, IBM, and smaller/specialized stuff (Excelstor, Toshiba for
Laptop drives))...

If Seagate would at least label their former Samsung drives in a
recognizable manner (say, ST*DS* vs. ST*DM* or keep the Samsung label
or whatever), I'd be a happy bunny, but as of now, that failing
ST3000DM001 was the last Seagate I've bought for quite some time.

Oh, and I've got a second ST3000DM001 used externally, with stuff that
can get lost, but it'd be inconvenient. And yes, I'm planning on
ordering a replacement ASAP (another WD40EFRX). Just in case. And the
steady state of my discs is full anyway ...

# dfall -t ext3 -t ext4 -h
[..]
14.6T        13.2T       996.4G  90%

and that's just because item one: /dev/sda is taken up by a 128G
Samsung 830 SSD, and item two: I just swapped in that 4TB WD for the
failing Seagate 3TB. (and yes, I use ext{3,4} exclusively on disk),
there's another ~10 TiB in the "fileserver" on 11 disks and a few
naked drives (and a docking station (Sharkoon QuickDeck)) and an
external drive ;)

>>> I ordered a drive.  It should be here tomorrow.  In the meantime, I
>>> shutdown and re-seated all the cables, power too. I got the test running
>>> again but results is a few hours off yet.  It did pass the short test
>>> tho.  I'm not sure that it means much. 
>>
>> Good. Do not use dd, it WILL fail at the first error. Use gnu ddrescue
>> or dd_rescue to grab an image. I used mc to copy via filesystem, eg. 
>> 'rsync -auxlPRAXSHD /foo/ /bar/' is fine too. Oh, and I hope you
>> didn't buy a Seagate again ;)

BTW: *GRR* for you buying a ST3000DM001 again

>I plan to rsync or cp the data over.

Good plan.

>The dd part will come into play after I am sure I got everything off
>that I can get and am just erasing the drive completely. I plan to dd
>the drive then run the tests again just to see what it is doing. 
>Heck, maybe it will reallocate that area like it should be doing
>already, I guess.

Reallocation happens on writes ... Look for --write-sector in 'man
hdparm'.

I've been doing that "smartctl -t .." / hdparm --write-sector ... 
stuff for a bunch of sectors but got tired of that game. And you
having 104 pending sectors? *gah* That'll get tedious. Probably using
'ddrescue /dev/zero /dev/sdX sdX.log' would be easier. That way, after
you got whatever data you can rescue from that drive, you can clear
the drive too before sending it in for a warranty replacement (if
still applicable).

>Time will tell.  I'll be having fun tomorrow tho.  ;-)

Do have fun, after you got your data off that drive :)

-dnh

-- 
Gibt es ein Buch über das maßvolle Verwenden von Fußnoten?
Wenn ja, dann bin ich bereit, Dir ein Exemplar zu schicken.
            [Thorsten Haude zu David Haller in sl-etikette]