HGST (Western Digital) 0S03664 Deskstar NAS H3IKNAS600012872SN (0S03839) 6TB 7200 RPM SATA 6.0Gb/s 3.5" High-Performance Hard Drive for Desktop NAS Systems (Retail Kit) for $300 - $30 off with coupon code ESCAKKS24 [Exp 1/17] = $270 with free shipping.

 

http://www.newegg.com/Product/Product.aspx?Item=N82E16822145973&nm_mc=AFC-C8Junction&cm_mmc=AFC-C8Junction-_-na-_-na-_-na&cm_sp=&AID=10446076&PID=404255&SID=407077_desktop

 

Deal is active again with w/ promo code AFEX0119, ends 1/25

I have two of these drives. I'm very happy with them.

They get about 225MB/s on the outer tracks.

Makes a nice parity drive. A NICE parity drive. 

how warm?

how warm?

 

It really depends on your case and fans.

on one of the drives it reached 45 after a week of 4 pass write/read of badblocks and 2 smart long tests. 

 

The drives are rated up to 60c

Environmental (operating)

Ambient temperature 5o to 60o C

 

http://www.hgst.com/tech/techlib.nsf/techdocs/0BF8B842F66D454288257C74007E842B/$file/DS_NAS_spec.pdf

 

Even in some Seagate FAQ's I've read, it stated that the larger drives are rated higher and the new norm is up to 60c.

 

The Seagates reached up to 42

However everything is now in the Fractal Design R4 and running in the higher 30s.

how warm?

 

It really depends on your case and fans.

on one of the drives it reached 45 after a week of 4 pass write/read of badblocks and 2 smart long tests. 

 

The drives are rated up to 60c

Environmental (operating)

Ambient temperature 5o to 60o C

 

http://www.hgst.com/tech/techlib.nsf/techdocs/0BF8B842F66D454288257C74007E842B/$file/DS_NAS_spec.pdf

 

Even in some Seagate FAQ's I've read, it stated that the larger drives are rated higher and the new norm is up to 60c.

 

The Seagates reached up to 42

However everything is now in the Fractal Design R4 and running in the higher 30s.

 

Am i under the impression that the only thing preclear does is zero out and read?

Am i under the impression that the only thing preclear does is zero out and read?

 

For Joe L's Preclear,  read, write, read again. write signature

 

 

My version of clearing(burning in) a drive

conveyance test (check log)

smart long test (full drive read scan) (check and save log)

badblocks, writes a pattern, reads it this occurs with 4 patterns.

x55,xaa,xff,x00

smart long test (full drive read scan) (check and save log)

 

That's 10 complete passes from start to finish of the drive. it takes upwards of 80 hours to finish.

This is the HGST estimate for a pass from smartctl. I think it finishes a bit faster.

This says it should take approximately 10 hours or so for a full pass of the drive from start to finish.

Extended self-test routine recommended polling time:        ( 632) minutes.

 

This is the Seagate 6TB estimate from smartctl

Extended self-test routine recommended polling time:        ( 652) minutes.

 

If someone could post the data for a 6TB RED we would have a full comparison.

hmmm

 

I use :

badblocks -c 10240 -s -w -t random -v /dev/sda

 

takes about 18 hours on a 3tb drive.

 

Guess i dont flip my bits as much as you:)  Still on the new shingle drives i hear you only get 69 flips per bit for the life of the drive...

WeeboTech's burn-in is probably a bit more thorough than the pre-clear script, but the key is to do SOMETHING that gives the drive a good workout before you put it in service.

 

I just use WD's Data Lifeguard utility -- I do the following:

=> Run a Quick Test

=> Run an Extended Test

=> Do a full write zeroes pass

=> Repeat the Quick and Extended tests

 

Then I do a Level 2 run with Spinrite

 

If it's a drive I'm going to add to an UnRAID server, I'll then do a quick run of the pre-clear script (skipping the pre and post reads) so it's cleared before adding it.

 

Still on the new shingle drives i hear you only get 69 flips per bit for the life of the drive...

 

I'm sure this is NOT true 

 

The magnetic functionality of the drives isn't any different than standard PMR drives -- there's simply some track overlap which requires re-writing the entire block of shingles when you've made changes.

 

Also note that WeeboTech uses x00 for the final pattern, so the drive ends up cleared. All that's needed then is write the clear signature.

Also note that WeeboTech uses x00 for the final pattern, so the drive ends up cleared. All that's needed then is write the clear signature.

 

True ... if you know exactly what needs to be written & where.    Clearly Joe knows what to write -- he and Tom discussed this years ago when the pre-clear script was first written -- but not sure if it's documented elsewhere (I don't recall having seen it here anywhere).      A simple pre-clear pass that skips pre- and post- reads seems simple enough to prepare a drive for adding to the array once it's been otherwise thoroughly tested.

 

My concern is always reliability of the drive over an extended period of time.

This is why shifting bit patterns are used. I remember reading all sorts of stuff about this with spinrite in the early days.

 

Truth be told, after my badblocks test. I stop the array, add it, recalculate parity.

Eventually I'll look at writing a tool just to add the precleared signature.

For now, I'm usually confident the drive will perform or die "$0: infant mortality" ; 

 

hmmm

I use :

badblocks -c 10240 -s -w -t random -v /dev/sda

takes about 18 hours on a 3tb drive.

Guess i dont flip my bits as much as you:)  Still on the new shingle drives i hear you only get 69 flips per bit for the life of the drive...

 

The issue here is the use of /dev/random takes up cpu cycles, thus slowing it down 'slightly'

Plus I'm not sure this is a full write, then read of the whole drive in a sequential manner.

 

A 4tb HGST 7200 RPM drive takes about 8 hours per pass for write, then 8 hours for read.

So 18 hours is probably appropriate given use of /dev/random and the speed potential for the drive.

 

The normal badblocks write mode test, writes the whole drive, then turns around and reads it. (Just like Joe L's)

This insures that data is not cached anywhere.

 

With random and/or read-write mode. data is written, then the same data is read back (I believe I have to check the code)

With read-write mode it's really even slower as data is read, stored, random data written, read, verified then the old data is written again. This DOES re-allocate pending sectors, but takes forever!

 

My 4 pass modified badblocks on a ST4000VN000-1H4168

shows this at the end

...
Jan  7 13:14:45 unRAIDb badblocks[32378]:  99.89% done, 63:24:47 elapsed. (0/0/0 errors), 7:54:00 elapsed this stage.
Jan  7 13:15:37 unRAIDb badblocks[32378]: done                                                 
Jan  7 13:15:37 unRAIDb badblocks[32378]: Pass completed, 0 bad blocks found. (0/0/0 errors)

They've extended the deal until 01/19, but changed the code and now it's $280

Deal is active again with w/ promo code AFEX0119, ends 1/25

This code is $34 off instead of $30, so drives cost $266

This code is $34 off instead of $30, so drives cost $266

Interesting... to me it's $30, just checked again...

This code is $34 off instead of $30, so drives cost $266

Interesting... to me it's $30, just checked again...

 

You are right, it is $30 off the HGST 6T NAS drives.

 

The reason for the extra $4 is I had one of these Rosewill 4in3s in my cart. Didn't realize that the same coupon code gave 10% off of it as well.

 

By the way, I can't say I'd give away my Supermicro 5in3s for these, but a 4in3 for $35 isn't bad. It used 120mm fans which is great for cooling! Mostly plastic, though. I actually use these as external enclosures for my backup array, and run a long SAS cable and a power supply lead to each one. Convenient package for 4 external drives. And since it is 4 and not 5, a single SAS cable works perfectly for one of these. A pair of these took my Antec 900 backup server case from 15 to 23 drives!