Announcement

Collapse
No announcement yet.

KDE Plasma 5.20 Will Alert You If Your Disk Is Failing

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • #21
    Originally posted by chuckula View Post

    He never said they were bullet proof, in fact he said the opposite. He said that when they DO fail they go from working fine to dead in an instant.

    And I've had multiple SSDs die exactly like that so I know he's right.
    This is why I am so scared to buy an SSD...

    Comment


    • #22
      Originally posted by Danny3 View Post
      Wow, that's a really cool feature to have for the unfortunate times.
      I wonder if it Linux filesystems have have something similar to Windows bad sectors relocation and if KDE Plasma can show that count too.
      I had only once a had drive failing and this was the sign which I discovered by using a third part too like CrystalDiskInfo which showed a high number of bad / relocated sectors.

      The good part was that it still had warranty and they replaced it without headaches.

      If this feature can warn users about this long enough before they lose the warranty, Plasma can even make people save money, which is really awesome!

      Congrats and many thanks to the developer of this feature and of course to the rest of KDE team for all the help!
      Well, I know ext filesystems can map out badblocks, but that is not something you should be doing on modern (say less than 20 years old?) hardware. Drives these days are meant to operate as black boxes presenting a linear array of sectors to the world. They contain a reserve of sectors that are transparently used to replace bad ones, either existing from manufacture, or as they crop up over time. This is generally done on write, or marginal reads (?) but hard read failure sectors are left alone, but accounted for in the SMART data. If you start seeing these failed sectors you should probably replace the disk. However you can force the drive to remap these sectors by writing to them with various techniques (this is what the various utilities will do if you ask them to repair the drive). This might work for a while, as long as the set of reserve sectors holds up, and the drive is not actually dying. You could also for example, use e2fsck to map out bad blocks in the filesystem, (and see them with dumpe2fs) but I would consider that a delaying tactic.
      Keeping an eye on SMART data, and doing periodic full surface scans (or enabling offline scanning on drives that support it) is probably a good idea, but as always the best thing is to make sure you have some sort of backup for the things you care about. Because drives can and do destroy your data and fail without warning.

      Comment


      • #23
        It used to be that Linux was an operating system, with or without a desktop environment. Now we have a strange announcement, that Linux cannot handle any hard disk failing, unless it has the KDE desktop environment. What about all those Linux operating systems, including Android, which do not have KDE? All Linux operating systems do not warn about failing disk systems, except KDE?
        If I dare rely on my 40+ years with computers, all computers can signal that storage is about to fail. When & how the operators are notified can vary. Strange that here it depends on the Desktop Environment (KDE) is now needed to do this function.

        Comment


        • #24
          I don't need a program for that. Boot problems, read problems, programs crashing...

          Comment


          • #25
            I'd say the vast majority of users don't realize they should use something like gsmartctl or "smartctl -a" to look at the health of their spinning media/SSD's. If you look at it periodically you can check if sectors are being reallocated on spinning media and on SSD's you can check wear leveling count. Both of those don't trigger SMART failures until they exceed vendor specific thresholds. Pro-actively monitor SMART values and take action before SMART indicates a catastrophic failure.

            Comment


            • #26
              Originally posted by chuckula View Post
              Oh, and "OMG YOU WROTE TO IT TOO MANY TIMES" is basically NEVER the reason for a modern [read: From the last 10 years] SSD to actually fail. It's always going to be something else unless you literally work in a lab that tests these things.
              If you wrote to a SSD too many times you would see the fail as a gradual degradation of the disk that would show up in SMART data because the remapping numbers and bad blocks would go way up. Fails that happen suddenly are due to something like the controller chip or the chip talking to the interface dying, not because you wore out the memory chips.

              Because of the speed the SSD drive can be written to and read from, it is possible to wear them out, but the amount of data that it requires is tremendous. The people with systems that are thinking they hit the drive hard and are worried about wearing out the drive. I can guarantee you are off on your data load by exponential proportions of what it takes to wear out a modern drive.

              Comment


              • #27
                Originally posted by tildearrow View Post

                This is why I am so scared to buy an SSD...
                If you want to go in gradually, put your personal data on a spinning drive and then let the SSD run the system software. You get many speed benefits of the SSD and the idea is that you can always re-install your Linux distro if things go south without losing anything important. Of course, backups are critical no matter what type of storage you choose anyway.

                Comment


                • #28
                  Originally posted by vsteel View Post
                  Because of the speed the SSD drive can be written to and read from, it is possible to wear them out, but the amount of data that it requires is tremendous. The people with systems that are thinking they hit the drive hard and are worried about wearing out the drive. I can guarantee you are off on your data load by exponential proportions of what it takes to wear out a modern drive.
                  I think the last time i checked, it was on the order of maximum write speeds to the drive 24x7 for 2-3 years straight in order to hit any kind of wear limitations. With the more expensive drives lasting longer.

                  Comment


                  • #29
                    I'm always a little worried about SSDs and I try to keep near-daily backups of any sensitive information on them, as I have had drives die with no warning. They always seem to just keel over instantly. I've never had one go slowly.

                    Originally posted by ssokolow View Post
                    I specifically double-checked that WD hadn't SMRed the 10 TB WD Reds before purchase and Seagate has publicly stated that they don't use SMR in their Ironwolf and Ironwolf Pro lines after those class-action lawsuits got filed against WD over putting SMR into the WD Red line. (I'd have considered going Toshiba+Seagate, but reviewers say comparable Toshiba drives are loud so I'll probably save that for when I can afford to use WD+Seagate+Toshiba to do something like Z-RAID + nightly online backups in a NAS stuffed somewhere I won't hear it.)
                    Seagate did something similar to WD with the 8TB Barracuda Compute drives; initially CMR, they snuck in a "refresh" which was SMR. I bought four at work after happily using a pair the year before - and was spitting fire when I saw classic SMR behaviour out of them. They obfuscate the hell out of the technical documents too, desperate to avoid calling SMR drives SMR because by now everyone knows they're rubbish.

                    ...

                    As for Toshiba drives, I'm (well, my department is) running dozens of the 12TB M07 Enterprise drives at work; they're some of the quietest drives I've used and they're really fast at both reading and writing, as well. The Ironwolf (Pro) and Toshiba Enterprise drives are my go-to disks for well behaved high capacity, although I favour the Toshiba drives simply because we can get them cheaper.

                    Comment


                    • #30
                      Originally posted by Paradigm Shifter View Post
                      I'm always a little worried about SSDs and I try to keep near-daily backups of any sensitive information on them, as I have had drives die with no warning. They always seem to just keel over instantly. I've never had one go slowly.


                      Seagate did something similar to WD with the 8TB Barracuda Compute drives; initially CMR, they snuck in a "refresh" which was SMR. I bought four at work after happily using a pair the year before - and was spitting fire when I saw classic SMR behaviour out of them. They obfuscate the hell out of the technical documents too, desperate to avoid calling SMR drives SMR because by now everyone knows they're rubbish.

                      ...

                      As for Toshiba drives, I'm (well, my department is) running dozens of the 12TB M07 Enterprise drives at work; they're some of the quietest drives I've used and they're really fast at both reading and writing, as well. The Ironwolf (Pro) and Toshiba Enterprise drives are my go-to disks for well behaved high capacity, although I favour the Toshiba drives simply because we can get them cheaper.

                      Seagate at least updated all their spec sheets (at least for still supported drive models) to identify SMR/CMR drives. They make the abbreviated spec sheets readily available on their site. WD still does not make such info explicitly available, last I checked.

                      Comment

                      Working...
                      X