18TB Enterprise HDD $424.99

TrumanHW

Member
Sep 16, 2018
114
8
18
Ha! I just mentioned these issues (1-4) to a client who'd mentioned wanting the 'biggest drive' possible (and looked to Amazon).



1. ...use Backblaze to infer the mfr's who take failure rates more seriously than profit-margins.

2. Amazon SANCTIONS the FLAGRANT sale of USED & REFURB HDs with little
(if ANY) warranty OR warning except via reviews! Ignore reviews at your HD-longevity's peril.

3. Remember, all drives ≤ 8TB are Helium, which means the ONLY thing we (data recovery) can do is disable heads, etc.

5a. Ace Labs (PC-3000) has STILL to add complete ANY support for Helium drives (despite ≤ 4 years on the market).

5b. Ace Labs (PC-3000) is $12k for their recovery platform & $3K annually for updates -- & is THE BEST recovery platform -- and even THEY have ZERO Helium support!.
6. I've received so many ≤ 8TB Seagates (Helium), to have recovered 4 IDENTICAL MODELS! (shucked 8s, I believe).
7a. I've received 3x ≤8TB (Helium) HGST drives, of which one's PCB was bad (12v TVS burned to to using the wrong adapter).
7b. Many HD failures due to using wrong PSU (power fault) use TVS for protection; thus, removing the diode + diode fuse provides imaging-access.
7c. Properly repairing the HD requires a good diode + fuse, to restore the protection against future (common) errors, especially if it's expensive & OOW!
7d. Oh, it's under warranty? Great! I'm sure vendors can't check a TVS & other indices which prove misuse to cite the limitations of their 'Limited Warranty'.
:cool:
7e. Fried your TVS? 2.5" HD only have 5v DC-in & 3.5" have both 5v + 12v DC in. (Shown on any 15p SATA power diagram). I have BAD eyesight, see below for more*
7f. As ID-ing diodes is hard!, it's easier to swap from VERY similar looking PCB of the same mfr. Swapping the PCBs CANNOT WORK!
(more on recovery below)
D.R. DD Rescue is a great FREE $2 tool to recover all drives (even RAID, esp repaired). It'll image the entire drive. You need an empty ≤ capacity HD. (more on recovery below)
8. Aside from PCB repairs & disabling heads (limits the size of recoverable files) -- to say the recovery options for Helium are limited is an UNDERSTATEMENT!!
9a. Model-endemic FW issues pose serious threats to an array (SA issues = less common), causing multiple failures after the SAME amount of TBW / hours used.
9b. It'd be truly disastrous for Flash arrays, as replacing PCB / fixing SA on spinning drives is possible. Perhaps homogeneous arrays pose a risk.
9b. While I also have the Flash recovery add-on, (which is also able to address SSD with failed FW if the drive is within the database, here's what you're talking about:
(see pictures re: Flash Recovery ... which doesn't even include the better software interface -- all of which is culturally done in Russian or Ukrainian -- which means the vernacular you've used for things will be NON-EXISTENT ... and you have no idea how much that accounts for your ability to learn something. I'm GOING to include a screenshot from PC-3000 ... so you can appreciate what I'm talking about here; pick a phrase that you want to understand, and I'll show you the support / definitions you're provided....(GOOD EFFING LUCK!))


I write a MINUTE intro to Data Recovery in case anything could be of use to anyone, & will post in a dedicated thread if deemed useful.


ARRAYS COMPRISED of DRIVES 8TB ...HAVE MORE REASONS TO USE DOUBLE PARITY (RAID-6 or RAIDz2).

ALL BELOW reasons are in addition to this standard caveat: "RAID / ZFS" are fault tolerant, but fault tolerant isn't a backup. True. But incomplete.
1. There are multiple new reasons: 8TB are homogeneously Helium filled, meaning there may be ONE or TWO companies WORLDWIDE that can open them!
2. The FAR FEWER recovery options for ≤ 8TB (Helium HD) effect on competition (far less) means YOU CAN COUNT ON MUCH MORE EXPENSIVE OPTIONS, IF ANY!
3. Ace Labs, without a doubt, THE BEST mfr of Data Recovery hardware planet-wide..? Has STILL yet to add a SINGLE Helium drive to their database.
4. We're talking about a company from whom my annual updates (yes, SOFTWARE UPDATES, ONLY) is $3,200. I have partial support for 2 Helium drives.
5. IF you're lucky enough that someone recovered the high-capacity drives comprising your array..? Ha! Wait till you figure out how RAID recovery prices are calculated. ;-)






DATA RECOVERY POINTERS:

A drive which makes no attempt to spin, doesn't hum, nothing. Just sounds like it has no power ... and if it's your drive, and you plugged in the wrong PS (POWER SUPPLY abbrev. from here out) bc many are similar in size (and why DIDN'T these asshats label their power supply with the make+model for which it corresponds..?? it's not like anyone has less than 20 or 30 PS, that're all black and give little indices to even it's technical specifications! But, once you smelled that sad smell of electronics burning, that tells anyone their voltage or polarity was wrong, chances are, it's gonna cost you. And sadly, the consequences of our errors aren't obliged to limit themselves to the degree of our errors ... (yes, of COURSE I've done this).

TVS, AKA, Transient Voltage Suppressors:

This happens pretty often to hard drives, as they seem to be rather sensitive to the voltage in; if you're doing this for a yourself or a friend, you can get away with removing the TVS; if the drive resumes working, just know that future errors will not be so cheap. If you're doing this for a customer, and you're charged ONLY with getting the data, you can do the same thing. If they'd like the drive repaired, it'd be unethical to remove a layer of defense to a common mistake made by those who regard voltage, amps, polarity and AC / DC as Enigma 2.0. ... because they'll do it again, and then, you'll have to replace the PCB AND -- MUST swap the ROM from their original hard drive.

WHY CAN'T I JUST REPLACE THE PCB..?

You know how cheap manufactures are..? Apple often won't put the 10-penny! (IF THAT) SMD component, the PCIe SSD interface on their 'Logic Boards' (motherboards) if the model doesn't include it at the point of sale. Yeah, you can solder one on later, for those who can. So, it's safe to say the ROM that's on the PCB has data on it which you'll need; namely, the P-List (permanent list of defects found during their initial scan of the platters) and the G-List (the 'GROWN list') ... while programmers understand how severe the "off by one" errors are..? (Again, things in which the error and consequence are completely disproportionate) ... imagine you're doing payroll on a spreadsheet which is paid out in alphabetical order. Now imagine that the columns are misaligned by one. How many people will be happy? For those of you in your mid-40s+ ... do you remember manually entering that list of errors on the hard drives cover..? Along with it's cylinders, platters, etc..? Right, that was when hard drives used the physical facts of define the sectors of the drive. But now, we have 'LOGICAL' partitions ... and the errors are now skipped whilst retaining sequential identity of each sector.

So yes, you'll need to swap the ROM
(unless...you have the equipment & know how to COPY the ROM, but if you...wait; are you just reading this to find my errors!??)



TVS DIODES & DIODES, GENERALLY:
Mfrs are assholes. They'll actually use ambiguous information if they even label their SMD components. Further, it's complicated to try testing components 'in situ' (without desoldering them) as you're often testing an entire circuit. Continuity you can test fortunately, and if nothing else, diodes should provide very different readings pending on which side the positive lead is testing. And unless there's an application I'm unaware of, there's going to be a direction in which the diode's resistance is very high.

Since HD schematics are a true rarity, and in only the scenario of TVS diodes (or other very similar things which I don't even know of) it's easier to either replace the PCB & swap the ROM, or, especially with manufactures like Hitachi (whose PCBs even have similar shapes for periods lately, along with the layout (despite the PCBs being utterly incompatible, as in, using totally different CPUs for instance) ... their TVS diodes (per voltage) are almost certainly interchangeable.

But also remember that if a diode is bad, there's at least a 50-50 that the fuse next to it (which can be freaking TINY) is also likely bad. If you get continuity to ground..? It's bad! If after removing both items the drive now powers up ... pending on your exact situation (recovering the drive vs repairing it) ... you can decide if you're going to replace it with a known-good diode and fuse. Because if removing both work ..? Replacing them with good ones will also (if done properly; obviously the diode's orientation matters!).

If it was a RAID drive ... were I you? I'd absolutely clone that sucker, ASAP!

If it's not part of a RAID array, a lot of other variables have to be weighed out; what's the drives health, is it going to degrade while copying it..? If so, and if I have access to a structure (either the HFS catalog is readable or the MFT is ... I'd pick the most important files, then folders ... this means asking the customer if it's not yours...especially if you lack the means to control how each file is treated during the recovery (skipped quickly and re-read more tenaciously after all the easily read files are recovered basically).


RECOVERING a RAID VOLUME:
• DD Rescue (macOS, Ubuntu, Windows) ...is a very good FREE imager ... or, if you have a linux machine:
• HDDSuperClone (requires Ubuntu / Linux) probably better but I've not used it, which is bc I own expensive imaging tools.
I only use DDRescue on the good drives of an array while recovering the actual patient drive.



WHAT DOESN'T WORK:
Apps like Disk Drill, Data Rescue EaseUS, and ~97% of Data Recovery sw are crap!, & fail to preserve the hierarchy / directory structure even when available. Each says they're better than the other -- and EVERY SINGLE APP I've SEEN PAY TO ADVERTISE IS GARBAGE. You could literally do what they do (just slower). They just read the raw content and look for headers and footers at the beginning and end of a range of HEX data that signify the filetype & boundaries.

JPG header: DD F8
JPG footer : DD F9
Copy from the first D to the last 9 to a hex editor, save the file as a jpg; if it was in-tact (not corrupt) you recovered the image.

SELECTIVE recovery is obviously the better choice when a drive is dying, (eg, your Home Folder, excluding your library, porn, etc.)
Even a small library will be very slow if it contains your Mail, iMessages, Browser Cache, etc., any small IOPS demanding files should be omitted if possible.
(during object-level recovery, of course) Use successive passes based on how hard it makes the drive work and importance.

FEW apps retain the structure of a macOS recovery. A lost hierarchy is also called a "RAW" recovery.
Raw recoveries are more common with general / small computer businesses which lack the knowledge & certainly equipment for better result.
(my cheapest hardware was ~$3000, my current most expensive costs $3,200 ANNUALLY just for UPDATES, the hardware over $16k, & I need another $14k unit)

Obviously imaging a drive via DD Rescue would retain everything it can image, if RAID, this is fine, but you'd still be at the mercy of the logical recovery sw.
It's entirely normal to image a drive to get the data -- and if it lost sectors before being imaged, it'll still require the use of logical recovery,
Hardware imagers like PC-3000 provide both object level (selective recovery) via multiple passes of bit-level processing ... It allows granular control over time outs, if this then that rules, etc., for each pass, never reading the same sector twice, can fix the Service Area (which isn't even ACCESSIBLE by 98% of expensive HARDWARE recovery platforms) ... and we've not even began to scratch the surface of what it can do, including RAID, 4 drives at once, etc. etc.
 
Last edited: