Barracuda XT, Barracuda, Barracuda Green, DiamondMax and other desktop SATA and IDE drives

[hyram]

I've been reading several posting here and elsewhere in search of a few clues concering the RAID drop outs I've seen as well as others are reporting. In my searching I ran across similar issues concerning WD desktop drives that have a longer error correction time out than the RE drives. WD refers to this as Time Limited Error Recovery (TLER) and this basicly disables the drive error correction mechanism allowing the RAID controller to do it's job... after all, this is what a good RAID controller should be designed to do. In the WD world there is a program (wdtler.exe) that can be found that can enable the TLER function that allows desktop drives be used in RAIDs, and disables it on RE drives so they can be used as regular desktop storage.

In the Seagate world this function is called Error Recovery Correction (ERC). What I don't know is if there is a program that will enable or disable this function. Also, I've found that there seems to be a way using the serial command connection to the radio (discussed elsewhere here) that should let one adjust the read and write error time outs. This is the limit of what I know. Does anyone know if there is a program to enable/diable the ERC function??? If not, has anyone with more experiance than I have tried to adjust the rea/write time outs usning the serial command connection??? Is this a difference between the AS and NS drives???

[HughR]

I'd like to know the answers to these questions too.  I've been asking some of them.

My current guess (very much a guess) is that these drives go offline not just for seconds but until a power cycle.  But there may be instances of both behaviours.  I've not gotten any clear reports from those who've experienced the problem(s).

I think you mean something like "serial port" instead of "radio".  As far as I know, that port is not intended to be used by customers.  The fact that some people have figured out how to use it is great from my standpoint, but probably very upsetting to Seagate.  I bet that Seagate would consider that using it would void your warranty.

Anyway, you won't get your questions answered by Seagate unless you ask them somewhere else.  Seagate does not post technical answers on this forum.  Sad but true.  You could try asking Seagate tech support.  If you do get an answer, please report back here.

[hyram]

HA!... "radio"... I ment "drive". That's what I get for trying to do more than one thing at a time.

I'm posing these questions here exactly because I wouldn't expect anything remotely like an answer from Seagate tech support. Hoping someone here has a few answers or at least a clue.

If not for the ERC... what IS the difference between an ST31000340AS and a ST31000340NS... other than the physical differences, ie: better berrings, more testing, etc??? What other firmware changes could there be???

[nick.lowe]

You need to issue the appropriate SMART commands to the drive to check ERC function availablity and to enable it. 

See http://www.t13.org/Documents/UploadedDocuments/docs2008/D1699r6a-ATA8-ACS.pdf

8.3.4 SCT Error Recovery Control command

Message Edited by nick.lowe on 10-01-2009 09:15 AM

[HughR]

nick.lowe:

http://www.t13.org/Documents/UploadedDocuments/docs2005/DT1701r5-SCT.pdf page 34 as counted by my PDF reader.

Very interesting!

Has this standard been passed?  Is this feature a mandatory or optional feature of the standard?

Do we know if these drives implement this feature?

Do the various OSes (Linux, *BSD, Solaris, OS/X, Vista, WinXP, DOS, ...) have provisions to issue this command when initializing a drive?  I don't remember hearing of any.  It would seem that the normal course of running start-up programs might be too late.

[nick.lowe]

I've checked with hdparm and can confirm that this feature is available on the Barracuda 7200.11, 7200.12 and LP series drives.

Also see http://forums.seagate.com/stx/board/message?board.id=ata_drives&thread.id=11830

Message Edited by nick.lowe on 10-01-2009 09:16 AM

[nick.lowe]

The difference between the consumer and enterprise class drives in this regard appears to be the initialisation state of this feature not its implementation.

If you do a hdparm -I device you can check if this feature is offered and its state.

Look for:

SCT Error Recovery Control (AC3)

...where present with an asterisk *, it's enabled:

* SCT Error Recovery Control (AC3)

I am not aware of a method of initialising drives with this feature. I will research this issue, however, as my interest is definitely piqued!

[m.j.tagged]

I've just bought two new ST32000542AS drives and I'm trying to use them to turn my 6 month old 2TB drive (also a ST32000542AS) into a RAID-5. Unfortunately my first attempt to build the RAID failed and the 6 month old drive was marked as the problem. A little research brought this thread up and a potential solution, which hopefully will solve my problem and yours.

The tool you need is part of smartmontools: smartctl version 5.4, which supports setting the SCT Error Recovery Control timeouts. There's also a patched version (5.39) available which is what I decided to use, as 5.4 hasn't made it into my Linux distro's packages yet and I wanted something working quickly:

http://www.csc.liv.ac.uk/~greg/projects/erc/

Smartmontools is cross platform, so this isn't a Linux only solution.

Here's what I did on each of my drives. 1) Check the current timings to see that they are indeed disabled:
# ./smartctl -l scterc /dev/sdc
 smartctl 5.39 2009-11-03 r2981 [i686-pc-linux-gnu] (local build)
 Copyright (C) 2002-9 by Bruce Allen, http://smartmontools.sourceforge.net

 SCT Error Recovery Control:
            Read: Disabled
           Write: Disabled

2) Set them to 7 seconds:
#./smartctl -l scterc,70,70 /dev/sdc
smartctl 5.39 2009-11-03 r2981 [i686-pc-linux-gnu] (local build)
Copyright (C) 2002-9 by Bruce Allen, http://smartmontools.sourceforge.net

SCT Error Recovery Control:
           Read:     70 (7.0 seconds)
          Write:     70 (7.0 seconds)

Obviously, repeated for each drive in the the array. Hope that helps a few people.

[teknik.fr]

Hello,

just to add a few informations to the last reply. mainly if you're using windows and/or 3ware cards

I didn't find any version of smartmontools with the command scterc enabled for windows AND 3ware drivers are not smart capable with smartmontools in windows.

So here is a workaround :

use ubuntu desktop cd 10.04 i386 to boot

install build-essentials (apt-get update then apt-get install buil-essentials)

download the good patched version of smartmontools , same address as above (http://www.csc.liv.ac.uk/~greg/projects/erc/)

extract (tar -zxvf sma....)

in the new folder sma....

./configure

make

make install

then you can look for your drives (for me

smartctl -i -d 3ware,0 /dev/twa0

smartctl -i -d 3ware,1 /dev/twa0

check the erc:

smartctl -l scterc -d 3ware,0 /dev/twa0

smartctl -l scterc -d 3ware,1 /dev/twa0

set it to 7 secs :

smartctl -l scterc,70,70 -d 3ware,0 /dev/twa0

smartctl -l scterc,70,70 -d 3ware,1 /dev/twa0

now you can reboot, remove the cd, and that's it.

BUT if you power of the computer changes will be lost so this workaround is ok for a server with an UPs, not really a solution for a private computer under windows.

I bought two barracuda LP 2TO 5900rpm. Before doing this I always had disconnections from the raid.

Now everything is fine. I transfered lots of datas to test, no disconnections.

Hope this helps.

[maxoptimus]

I had four ST32000641AS smartctl -l scterc reports drive don't support sct commands? somebody known were to get more info about these drives how work, support/unsupport AC3, etc. Maybe some updates to smartctl can enable this feature or not?