Drive Failed, considering hybrid - what to look out for?

[grahamperrin]

Various

Data recovery

I cloned her existing drive over to her external drive using Phoenix. Phoenix registered a lot of I/O errors …

Stellar Phoenix Windows Data Recovery, yes?

(There's also Phoenix Data Recovery but AFAICT at a glance they don't offer software solution for use by customers.)

Incidentally, for future reference: with Mac hardware you may find it more convenient (and maybe less risky) to boot Ultimate Boot CD then start Parted Magic to run ddrescue (very highly regarded data recovery software). By coincidence at the time of writing I have both (a) Parted Magic running ddrescue on a USB 3.0 drive attached to a relatively old MacBook Pro that's normally limited to Snow Leopard; and (b) ddrescue on OS X running on the 2009 MacBookPro5,2 that I'm using to make this post. (Side note: a query on ddrescue data transfer rates in the OS X environment.)

I say 'less risky' only because I have not looked to see whether the routines attempted by the Phoenix software are suitably orderly for media that's on its way out.

Solid state hybrid drives

… haven't seen any high praise for hybrid drives yet. It would be interesting to get some stats on their real performance.

Below …

See if you can still find a Seagate Momentus XT. These are 7200 RPM. The second version is 750GB. They are real self contained hybrids. …

Smart. I got one of those put in 2009 MacBookPro5,2 after the original HDD suddenly and completely failed (a known problem with that particular range of Hitachi drives).

Somewhat technical: concerning the Seagate solid state hybrid and checksum mismatches reported by fsck_cs, my December 2014 edition to http://apple.stackexchange.com/a/139639/8546 may be of interest. tl;dr a few months ago I began wondering whether the failures, which apparently began around that time, are at the hard disk level or solid state level, or both. (I continue to use the suspect drive only because I have an unusually strong approach to data integrity for my home directory and backups thereof.)

Off-topic: Mavericks in the App Store

… one star reviews for Mavericks greatly outnumbered the four and five star …

Related:

10.9 to be the last of the old school ? – App Store ratings for OS X Mavericks​

Solid state and sudden failures

… Although SSDs don't have mechanical parts there does seem to be some evidence that they can just lose everything all at once in the right circumstances. …

Yeah, I don't have the relevant bookmarks handy but I lost count of the number of times that I found discussion of sudden loss of everything with solid state. One link handy, a local example from just over a year ago:

"… switched his 6 month old MacBook Pro on to get the folder with a flashing question mark; the SSD had failed catastrophically. …"​

ZFS

… Time Machine to restore it. I'm not so certain that the girl I'm talking about takes that type of precaution.

BradHatter, this'll not be for your sister's friend but as you have an interest in storage … for the future, you might like to familiarise yourself a little with OpenZFS on OS X. It's too early for me to recommend it (I have not tested the pre-release that's intended to support Spotlight) but L2ARC can be a very cost effective way of adding performance .

 

[TheBSDGuy]

Yeah, I don't have the relevant bookmarks handy but I lost count of the number of times that I found discussion of sudden loss of everything with solid state. One link handy, a local example from just over a year ago:

"… switched his 6 month old MacBook Pro on to get the folder with a flashing question mark; the SSD had failed catastrophically. …"

A lot of those catastrophic failures are actually SSD resets where the entire data field is being nulled (zeroed) which is like an instantaneous format. The SSD is still usable. It's not as if all of a sudden every single RAM chip died.

 

[BradHatter]

@Graham Perrin:

The Phoenix that I'm using is the product that came with Scannerz. What it does is extract the core OS from a volume and create an e-drive so you can use it for testing. It can also do basic volume clones and will record what files are bad files as it clones. I underline "basic" because it is definitely not something like CCC or Super Duper. It's more oriented towards creating an e-drive and then getting contents off a bad drive or doing very basic cloning.

When I first got Scannerz I created what's called a Phoenix Boot Volume from my own system and put it onto a USB flash drive. You're supposed to do this so you can "unload" the internal drive if there's something wrong with it and it needs testing. Anyway, I remembered I had that and just plugged it into her system and that's how I tested it. I personally suspected initially that her logic board was on the way out because it would always crash on boot. I thought for sure that when I booted off the Phoenix drive it would immediately crash and then I could tell her to take it to Apple for further diagnostics or get a new system. Much to my surprise, it booted fine from the Phoenix drive, so I ran a Scannerz test on her system. The hard drive had a pretty significant amount of damage scattered around the 1G to 5G range, if I recall correctly. Scannerz diagnostics mode confirmed it was the drive and not a cable or other problem with the system.

I put the Hitachi into her system, put in a fresh copy of Mavericks and then imported the rest of her stuff from what I was able to get off the original drive. She wanted to try Yosemite but when I showed it to her on one of the external partitions I have with my own system, she had problems seeing it. She's hardly blind but I guess has some problems with stuff that's close and the new fonts combined with translucency on a dark background were apparently irritating to her.

In any case, everything seems to be working well now. The Hitachi is surprisingly fast for an old style hard drive.

@BSDGuy:

If there's one thing I've learned over the last few months about mass storage, whether SSD or old style hard drives, it's this:

TRUST NOTHING!

I was really surprised when my original SSD failed so quickly, but the replacement's working fine. My plan of action is to use the SSD for my internal storage and a good, big, fat old time mechanical hard drive to do the backups. I'll probably check both of them once a month by doing a scan on them and hopefully catch any problems before I end up with a dead drive.

I figure if my SSD experiences one of those total loss of data events I can just restore it. Let's just hope such experiences are oddities, not the rule for SSDs.

We'll see.

 

[TheBSDGuy]

I don't know if calling the Hitachi you installed an "old fashioned hard drive" is appropriate. It uses 4K sectors mapped to 512 as 512e, has a higher spin rate than most consumer drives, a larger buffer, and a much higher areal density which yields a much higher media to system data rate. Hitachi also has some enterprise class drives using the same sort of technology that are even faster.

That's just a comment/opinion, not a critique.

 

[grahamperrin]

Reusing an SSD from a failed SSD drive

BradHatter: thanks.

A lot of those catastrophic failures are actually SSD resets where the entire data field is being nulled (zeroed) which is like an instantaneous format. The SSD is still usable. It's not as if all of a sudden every single RAM chip died.

Does reusing the disk in a working drive require specialist tools or skills? (Just curious. I don't expect to do that, ever.)

 

[TheBSDGuy]

BradHatter: thanks.



Does reusing the disk in a working drive require specialist tools or skills? (Just curious. I don't expect to do that, ever.)

I'm not sure what you mean. If you're referring to an SSD that essentially resets itself, it's doing just that. It's like the drive was fully partitioned, formatted, and loaded with data, then after the incident, it's like it's brand new, out of the box, containing absolutely nothing. As you might guess, the problem isn't (or wasn't) that widespread, otherwise no one would be using SSDs.

 

[BradHatter]

There's a lot of negative stuff about SSDs on the web, but one thing that I've learned is not to take too many of these reports all that seriously. I had bad luck with my initial SSD but it was just that - bad luck. My new one works fine.

Remember people have problems with RAM chips, logic boards, video displays, etc. all the time. A consumer can't take these one-off reports or analyses all that serious without a real statistical base for comparison. Often that doesn't exist.

 

[ZVH]

Every hard drive I've owned has lasted at least three years without any problems. I haven't owned an SSD that long so I can't compare it to hard drive reliability. I would guess, or at least hope, that SSDs reliability will equal or exceed that of hard drives, but there is at least some evidence to suggest otherwise.

Only time will tell.

 

[BradHatter]

It seems to me that SSDs really didn't start getting used much until about 2012, which is really, considering it's very early 2015, just a little over 2 years ago. SSDs haven't been around long enough to stand the test of time, IMHO.

 

[yjchua95]

It seems to me that SSDs really didn't start getting used much until about 2012, which is really, considering it's very early 2015, just a little over 2 years ago. SSDs haven't been around long enough to stand the test of time, IMHO.

I've had a Samsung 830 512GB SSD since 2011 and it's still running fine at full speeds.

And relax, SSDs will last for quite a long time.

http://techreport.com/review/27436/the-ssd-endurance-experiment-two-freaking-petabytes

Especially the MLC ones. 2 petabytes of data have been written to it and it still hasn't failed.

 

[ZVH]

If I recall correctly, OCZ and Corsair had some problems a few years ago. I think it was OCZ that had the "give up the ghost" problem where it would just wipe the disk clean, and Corsair had a problem with not recording or recognizing bad blocks. That was probably three years ago.

I found the following list of SSD manufacturers, for those interested:

http://en.wikipedia.org/wiki/List_of_solid-state_drive_manufacturers

I found the name "Bigboy Technologies" sort of interesting. Wasn't/isn't "Bigboy" the name of a hamburger chain in the midwest?

 

[grahamperrin]

… an SSD that essentially resets itself, it's doing just that. It's like the drive was fully partitioned, formatted, and loaded with data, then after the incident, it's like it's brand new, out of the box, containing absolutely nothing. …

Thanks. I guess that in some such cases, some data may be recoverable.

(Exceptions – no chance of recovery – might include cases where an ATA Secure Erase command https://en.wikipedia.org/wiki/Solid-state_drive#Data_recovery_and_secure_deletion had occurred unexpectedly … true?)

It seems to me that SSDs really didn't start getting used much until about 2012 …

SSDs or SSHDs?

 

[TheBSDGuy]

Thanks. I guess that in some such cases, some data may be recoverable.

The reset I'm referring to is a firmware command that essentially marks all blocks as ready for erase, so I'm not sure it it's recoverable or not.

 

[BradHatter]

@grahamperrin:

I'm under the assumption that hybrid drives only came out a few years after SSDs did themselves. That's my opinion/suspicion, not a fact. I just really don't remember seeing such things on the market until, quite honestly last year.

 

[grahamperrin]

Checksum mismatches with an SSHD from around 2012

… I just really don't remember seeing such things on the market until, quite honestly last year.

I don't remember this PR, but it was released in May 2010: Seagate delivers World's Fastest Hard Drive for Laptop Computers; Momentus XT Solid State Hybrid Drives offer Solid State Performance with the Capacity and Value of Hard Disc Drives | Seagate

If I recall correctly, I chose a faster Momentus XT in Summer 2012 and signs of trouble became noticeable around two years later.

Not intended to be comprehensive or conclusive, just for the record:

2014-07-01 22:18:42 +0100: Invalid Disk Label @ 132809609216: cksum mismatch
2014-10-02 13:01:37 +0100: Invalid Disk Label @ 133167747072: cksum mismatch
2014-10-07 09:41:43 +0100: Invalid Disk Label @ 133176135680: cksum mismatch
2014-11-08 04:13:17 +0000: Invalid Disk Label @ 133163552768: cksum mismatch
2014-11-13 18:25:07 +0000: Invalid Disk Label @ 133163552768: cksum mismatch
2014-12-02 09:02:58 +0000: Invalid Disk Label @ 133171941376: cksum mismatch
2014-12-14 13:40:02 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:04:46 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:04:50 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:05:08 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:05:15 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:05:39 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-14 15:05:44 +0000: Invalid Disk Label @ 133176135680: cksum mismatch
2014-12-30 20:39:10 +0000: Invalid Disk Label @ 133163552768: cksum mismatch
2014-12-31 23:50:24 +0000: Invalid Disk Label @ 133171941376: cksum mismatch

A few hours ago:

Invalid Disk Label @ 4096: cksum mismatch​

– and this evening, for the first time with this drive, I can identify a particular file that has lost integrity.

(I'm not complaining – the notebook has been dropped a few times, and so on.)

 

[BradHatter]

@grahamperrin:

We went with a Hitachi because it was big enough, fast enough, and above, cheap enough, dirt cheap in fact.

What do you think is causing those errors? Is the SSD portion not syncing properly with the HD? Bad shutdowns? Like you said…drops? I'm under the impression that the hybrids use the SSD as sort of a super cache for the drive. I have to wonder what happens when the SSD section starts having all it's write cycles used up. Does it just act like a normal drive then and ignore the SSD?

There are a lot of strange combinations of things I could think of as possible failures with one of the them.

 

[yjchua95]

@grahamperrin:

We went with a Hitachi because it was big enough, fast enough, and above, cheap enough, dirt cheap in fact.

What do you think is causing those errors? Is the SSD portion not syncing properly with the HD? Bad shutdowns? Like you said…drops? I'm under the impression that the hybrids use the SSD as sort of a super cache for the drive. I have to wonder what happens when the SSD section starts having all it's write cycles used up. Does it just act like a normal drive then and ignore the SSD?

There are a lot of strange combinations of things I could think of as possible failures with one of the them.

My best bet is that OS X doesn't play nice with SSHDs, simply because it's not like a Fusion Drive (which has a dedicated SSD and separate HDD. OS X sees two separate drives but treats it as a single volume and intelligently shifts data between the two.)

In the case of SSHDs, OS X can't do the same, since it only sees a single drive. OS X will also conflict with the drive's algorithm and cause all sorts of things to happen.

 

[BradHatter]

My best bet is that OS X doesn't play nice with SSHDs, simply because it's not like a Fusion Drive (which has a dedicated SSD and separate HDD. OS X sees two separate drives but treats it as a single volume and intelligently shifts data between the two.)

In the case of SSHDs, OS X can't do the same, since it only sees a single drive. OS X will also conflict with the drive's algorithm and cause all sorts of things to happen.

I would think the firmware in the drive should isolate the OS from the drive internals, but that's a guess.

 

[TheBSDGuy]

The firmware, if done properly, should completely isolate the inner workings of the drive from the system. The firmware should present the hybrid to the system as a single hard drive. A fair number of more expensive RAID units do this sort of thing.

 

[grahamperrin]

Core Storage checksum mismatches, identifiable corruption of data

… What do you think is causing those errors? … Like you said…drops? …

Maybe.

Luckily the recently corrupted file in my home directory was something that was trashed. I have backups of corruption-free data from that area, and so on.

There might be unidentifiable corruption of data elsewhere on the disk – in the OS X startup volume, with apps and so on – but that inability to identify corruption is pretty much the norm with HFS Plus.

My best bet is that OS X doesn't play nice with SSHDs, simply because it's not like a Fusion Drive …

… OS X will also conflict with the drive's algorithm and cause all sorts of things to happen.

Interactions are fine. I'm not aware of any conflict.

… present the hybrid to the system as a single hard drive. …

Yep, with a rotational rate.

gpes3e-gjp4:~ gjp22$ date; sw_vers ; system_profiler SPSerialATADataType
Wed 14 Jan 2015 20:13:01 GMT
ProductName:	Mac OS X
ProductVersion:	10.9.5
BuildVersion:	13F34
SATA/SATA Express:

    NVidia MCP79 AHCI:

      Vendor: NVidia
      Product: MCP79 AHCI
      Link Speed: 3 Gigabit
      Negotiated Link Speed: 3 Gigabit
      Physical Interconnect: SATA
      Description: AHCI Version 1.20 Supported

        ST750LX003-1AC154:

          Capacity: 750.16 GB (750,156,374,016 bytes)
          Model: ST750LX003-1AC154                       
          Revision: SM12    
          Serial Number:             W2006N8C
          Native Command Queuing: Yes
          Queue Depth: 32
          Removable Media: No
          Detachable Drive: No
          BSD Name: disk0
          Rotational Rate: 7200
          Medium Type: Rotational
          Partition Map Type: GPT (GUID Partition Table)
          S.M.A.R.T. status: Verified
          Volumes:
            EFI:
              Capacity: 209.7 MB (209,715,200 bytes)
              BSD Name: disk0s1
              Content: EFI
            disk0s2:
              Capacity: 615.98 GB (615,982,063,616 bytes)
              BSD Name: disk0s2
              Content: Apple_CoreStorage
            Boot OS X:
              Capacity: 134.2 MB (134,217,728 bytes)
              BSD Name: disk0s3
              Content: Apple_Boot
              Volume UUID: 67280920-1D8C-3471-BBB3-D5A3FDD44813
            disk0s4:
              Capacity: 133.18 GB (133,180,334,080 bytes)
              BSD Name: disk0s4
              Content: Apple_CoreStorage
            Recovery HD:
              Capacity: 650 MB (650,002,432 bytes)
              BSD Name: disk0s5
              Content: Apple_Boot
              Volume UUID: 994475C3-6F77-3F2C-BE10-806393A87345

    NVidia MCP79 AHCI:

      Vendor: NVidia
      Product: MCP79 AHCI
      Link Speed: 3 Gigabit
      Negotiated Link Speed: 1.5 Gigabit
      Physical Interconnect: SATA
      Description: AHCI Version 1.20 Supported

        HL-DT-ST DVDRW  GS21N:

          Model: HL-DT-ST DVDRW  GS21N                   
          Revision: SA18    
          Serial Number: KZ991SE2440         
          Native Command Queuing: No
          Detachable Drive: No
          Power Off: Yes
          Async Notification: No

gpes3e-gjp4:~ gjp22$

 

[BradHatter]

@grahamperrin:

Two question:

1. Since you have a hybrid, how would you rate it's performance compared to a regular hard drive. Noticably faster, moderately faster, slightly faster, or not significantly different from other hard drives (I'm excluding slower, because I don't think it's an option, but I could be wrong)

2. Why two core storage? The hybrid didn't do that did it?

 

[FrtzPeter]

@grahamperrin:

Two question:

1. Since you have a hybrid, how would you rate it's performance compared to a regular hard drive. Noticably faster, moderately faster, slightly faster, or not significantly different from other hard drives (I'm excluding slower, because I don't think it's an option, but I could be wrong)

2. Why two core storage? The hybrid didn't do that did it?

Core storage is the default for Yosemite. That's my guess.

 

[grahamperrin]

SSHD noticeably faster than HDD

Core storage is the default for Yosemite. That's my guess.

Core Storage is not a default for all Yosemite environments.

I don't use Yosemite, please see the ProductVersion part of my previous post.

… 1. Since you have a hybrid, how would you rate it's performance compared to a regular hard drive. …

Noticeably faster.

----

Off-topic from SSHD,

2. Why two core storage? …

Questions on FileVault and security: Not allowing other administrators to access my home directory

 

[ZVH]

Questions on FileVault and security: Not allowing other administrators to access my home directory

A trick I hadn't even thought of!

 

[TheBSDGuy]

It would be interesting to see a performance comparison beween a hybrid drive and one of the new AF Hitachi's, particularly some of their enterprise class drives.