Results 1 to 7 of 7

Thread: SSD with bad blocks

  1. #1
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default SSD with bad blocks

    Last year I bought a used 1TB SSD. Bad move. I partioned it and started using for various backups. Seemingly it works fine.
    BUT, when I tested it with utilities, all tests showed bad blocks. Now, I know that certain amount of bad blocis can be tolerated - the SSD comes with extra reserve of blocks, and when one attempts to write a file on bad block, a good block from the reserve is allocated for it. So, there I have a 1TB SSD hard drive, with almost 50% of space used, most of the files can be read and often I can copy to it. But a utility like Disk Warrior tells me "all spare blocks have been exhausted". And I was able to locate some files which can no longer be read. Obviously, this dirve will be junked because I heard of no way of fixing it.

    So, now I will have to buy a new one and obviously it will come from MacGurus. Ok, so I order one (1TB crucial) and do a full surface scan before I start using it. Most likely it will show some bad blocks but the reserve will be untouched. So I start using. Suppose in time, in a year or so, more bad blocks develop and I will come to near the end of the reserve. Will Rick replace it with a new one? I am sure he will, hopefully without any $ loss to him. But how do I return it? I would insist on total secure erase, perhaps triple pass with a utility like Drive Genius.

    Rick, is that the way to go in case of failure? And one more question: if a file is sitting on a bad block, can it really be wiped out?

    A curious observation: when I try to read the bad file from SnowLeopard, it tells me it is damaged and that's it. No problem. But when I try to read the same file from Sierra, most often the computer either freezes or shuts down, as if in kernel panic. But it restarts ok.
    marrand

  2. #2
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,122

    Default

    Well, I don't have any capability to warranty replace drives. Those always and for everyone get handled by the manufacturer. That includes spinning drives as well as SSDs. Any retailer has recourse to replace DOA drives in the first few weeks, after that, by warranty terms, all coverage goes to the manufacturer.

    I do not know what is proper for secure erasing an SSD. My method was to use several of them in a RAID, write a huge data file, the delete the RAID. Not an effective choice with a single drive. Backblaze has a decent article on the issue.

    I guess I don't sell many used drives. And if I send one back for warranty, it usually is a deader. Yeah, it could have my data on it. Once an SSD is dead though, which is a much more common type of failure on an SSD than a progressive slow failure like you are experiencing, there is nothing you can do to burn the data off it without destroying the drive.

    I guess I'm not much help here. There is no perfect answer. Maybe encryption is part of it.
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  3. #3
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Tongue

    Quote Originally Posted by ricks View Post
    I guess I'm not much help here. There is no perfect answer. Maybe encryption is part of it.
    Rick,
    You actually helped me a lot. Order for 1TB crucial SSD plus other items already placed.

    Regarding Secure Erase. The article by Cohen clearly states it is possible by zeroing out (1 or many passes) the entire drive from the Terminal. But there are utilities, such as Parted Magic or Drive Genius which do the same and easier. Some, however, prefer not resorting to this at all; instead use the File Vault. But that is mere encryption. The individual cells and blocks still contain your data, only the directory is scrambled. And then comes this fellow Daniel Howard, commenting on the article by Cohen, who maintains that an SSD can be disassembled, chips pulled out, and "without much effort" the fragements of "securly erased" data can be read. This is getting into heavy stuff which I have difficulty digesting.

    My simple man's way of knowing if a drive has been securly erased is to try to recover the erased data, using some state of the art utility capable of "deep" recover. I tried Disk Drill and it seemed to work well: recovers nicely all trashed files, but can't find any which were secure erased by overwriting everything with zeros. Will there still be readable fragments? Perhaps, but who is going to spend time dissassembling the SSD to find them!?

    The "bad blocks" is different issue. The one I mentioned needs to be replaced because some files on it already are "dead". I used TechTool to do complete surface scan. Result: 49,307 bad blocks enountered. Drive Genius shows 41, 000. DW says the spare good blocks have been exhuasted. (1TB SSD has a little over 2 billion blocks).

    And yet, junking this drive doesn't really make sense to me. If the number of blocks is constant (came thus from factory), the drive will function marvelously until the reserve of good blooks has been depleted. But how can that be? The drive is only half full; all unused blocks are available for the reserve, right? So why not use those? As I understand it, SSD internal management doesn't simply allow the space to be overwritten like on old spinning magnetic drives. It takes the new data and puts them in fresh reserved blocks, and only later does it return to the duty of cleaning up the garbage from those in which the data is no longer needed. An unused block does not mean it is ready for use; it must be erased first. This process is not automatic - the garbage disposal process doesn't know where the garbage is and just goes blindly along. And it is not always active (for best results, it must be kickstarted). The TRIM command from the OS will inform SSD management where this garbage is and thus the clean up can continue faster. I will try both on this crippled drive just to learn more about SSDs.

    Will the bad blocks increase in number? If so, nothing can help it. If not, then most of the disk may be usable.

    So Rick, it seems a bad SSD is judged NOT on how many bad blocks it was born with, but on how fast they are multiplying. When the new 1TB crucial arrives, I am sure it will be ok. I will rezero it first and do a full suface scan, to get the first read out on bad blocks. Then, if they multiply, I will have to send it to the manufacturer. Is that about right?

    See? Often you can help a customer without much effort.
    marrand

  4. #4
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,122

    Default

    John at the warehouse tells me he has Neutrino coming in already in a stock shipment for next day or so. And the Crucial is due a day later. So shipping to you will be in a couple days.

    I really like the Lycom USB3.1 cards. Just rebuilt a MacPro 4,1, installed 12 core 3.46 GHz processors, nVidia R680 graphics, Crucial 1TB MX300, and the same card as you just ordered. This rig comes so close to the performance of my 2013 cylinder MacPro as to be virtually the same.
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  5. #5
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Confused

    The stuff arrived arrived fine and is working just great. I will install the Lycom later (after taxes are done).

    But puzzle me this, oh wise macgurus. In the starting post I described a bad 1TB SSD - very many bad blocks. So many that some files could no longer be read. Yet the bootable partition booted ok. TechTool showed over 40,000 bad blocks, and DiskWarrior just said there are bad blocks. So, I got myself a new 1TB from Rick, and started working on it. First, I ran the full surface scan. No bad blocks. Then zeroed out the entire drive. Then partitioned and populated, and tested again. Neither TechTool nor DW showed bad blocks. That was a relief because I am using about 500 GB (for both partitions).

    And the bad SSD? Well, it will be destroyed but just for curiosity I secured erased it (multiple passes). Then tried to recover erased files with DiskDrill and there were none to recover. But....and here is the puzzle....this SSD no longer has any bad blocks. Surface scan by TechTool shows none, when before it recorded over 40K. What happened to them? Other utilities show no problems at all, as if it is perfectly healthy.

    But that's not all to this puzzle. I just couldn't believe that full secure erase can repair bad blocks. I am sure it can't. So, why none showing? Could it be that the surface scan shows ONLY the indexed bad blocks? When there are no files to index, the bad ones just don't appear on the list? Is that the explanation? Going on this theory, I checked DW again; yep, the same comment again: the disk has bad blocks. How does DW know when other utilities don't?

  6. #6
    Join Date
    Aug 2001
    Location
    Grangeville, ID USA
    Posts
    9,122

    Default

    All this begs the conclusion that disk utilities do not know how to analyze SSDs. Always said that most of them don't do spinning disks well. I guess the answer is they do just as well on SSDs.
    molṑn labe'
    "I am a mortal enemy to arbitrary government and unlimited power. I am naturally very jealous for the rights and liberties of my country, and the least encroachment of those invaluable privileges is apt to make my blood boil."
--Ben Franklin

  7. #7
    Join Date
    Jul 2001
    Location
    maryland
    Posts
    564

    Default

    Quote Originally Posted by ricks View Post
    All this begs the conclusion that disk utilities do not know how to analyze SSDs. Always said that most of them don't do spinning disks well.
    Yes, I remember your opinion of utilities from way back. But they may not be as bad as you think.
    .
    Take this sad SSD which I discussed above and destroyed. It had two volumes: smaller for booting into SnowLeopard, and the larger one for strage.
    From another SSD (Macgurus) inside my MacPro, which has the latest updated Sierra and updated utilities, I would occasionally check all the drives. And then one day they told me that this sad SSD has problems and about to fail. Nah, I said to myself, I won't believe you; Rick doesn't. For a while things ran fine, and finally the bootable volume no longer could boot. No utility, not even DiskWarrior could salvage it. Thinking it was merely a software problem, I erased that volume, reformatted, and cloned it again. (It is my backup anyway).

    All was running fine, yet the utilities keep saying this disk is about to die. So I ran the surface scan. Oh look at those bad blocks. Checked many files in the storage volume (same disk) and many couldn't be read. The bootable part kept booting, but some files gone bad too. You know the rest. I decided to destroy it and now am using the one I got from you.

    So, the utilities (DriveScope, DriveGenius and DiskWarrior) all warned me about the problem. The problem was there, and the volume did crash. Good thing is was only a backup. Yet, the details of the problem varied from one utility to another, and that was my question in last post.

    Crucial has its own utitlity, the manager or something, but for Windows only. I installed it on BootCamp Windows 10 partition. It is totally useless to me. Mac utilities tell me more. At least today, every disk I have, SSD or spinning, is in perfect order. (A few are over 10 years old).

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •