From: nospam@needed.invalid   
      
   On Thu, 2/19/2026 9:02 AM, Lars Poulsen wrote:   
   > Summarizing until now:   
   >   
   > My rebuild of my Linux home workstation/server with new Seagate 2TB   
   > drives has made me look for how to best reuse the older drives. A   
   > handful of 1TB drives and external HDD drives of 1TB, 2TB and 6TB.   
   > Some of these were recycled and I had been using them in open   
   > "caddies", while others were bought in external enclosures for use as   
   > backup drives. Since the triggering factor for the rebuild was   
   > discovering "bit rot" in my large archives of JPG photo files   
   > (approaching 2TB), and MP3 music files (about 600GB by now), that   
   > had been building up so gradually that when I found it, even some   
   > 2 year old backups were bad, I have now become acutely aware that   
   > I need to look at reliability issues and monitor SMART data so I get   
   > warnings before the errors become uncorrectable.   
   >   
   > Quite recently, I had bought a 6TB external drive "Seagate Backup+ Hub"   
   > with the intention of consolidating the video files that had been   
   > squeezed into corners with a little free space here and there into   
   > a single file system. As I was preparing to clean and repartition it   
   > for use with the new system, I discovered that Linux will not show   
   > SMART data for it. That has made me consider ripping the drive out of   
   > the enclosure and putting it in a "better" enclosure.   
   >   
   > There are a ton of enclosures (as well as caddies/adapters) on the   
   > market, starting at $10 and with many in the $18-$30 range. All ofr   
   > these are Chinese made, and even on Amazon, most of them are shipped   
   > from China. Whether on Amazon, Walmart or other marketplaces, there   
   > is no reliable technical information posted, such as whether they   
   > support SMART or what chip(set) they use. While I have good experience   
   > with caddies, they look messy on a desk, and scare the house cleaner,   
   > so I'd like a tidy enclosure. When looking at enclosure reviews on   
   > Amazon, less than 80% (in some cases down to 60%) have 5 star ratings.   
   > Some reviews say "dead on arrival", "computer did not recognize drive   
   > when mounted in enclosure" on other obvious flaws that would give you a   
   > chance to return the product, others describe initial success, followed   
   > by failures after a week, or 3-4 months, that seem consistent with   
   > overheating quickly but gradually deteriorating the controller board,   
   > often killing the drive.   
   >   
   > There are some enclosures where the supplier explicitly says that   
   > SMART is supported, the product is stocked in local Amazon warehouses   
   > and there is a decent warranty, but now we get into a different price   
   > range: $60-$80. At which point, why bother: For not a lot more, I can   
   > buy a new external drive package from Seagate or Westerni Digital with   
   > a decent warranty (2 years with my BestBuy membership).   
   >   
   > But it bothers me to throw away drives, some of which are likely   
   > to have years of service life in them.   
   >   
   > It looks like Paul has similar thinking, and has put efforts into   
   > researching the topic.   
      
   My method would be to test the drives while SATA connected.   
   That ensures that with a modern SMART tool (not HDTune 2.55),   
   the definitions of the SMART entries is known.   
      
   SMART does not guarantee that trouble can be spotted. I've also   
   used read benchmarks as a metric. But this is very difficult to   
   do in a convincing manner -- quiet scans are tough to set up.   
      
   Using a Win10 32-bit OS installation, then installing Macrium 32-bit   
   version, then making 32-bit Rescue Media, you can boot from a   
   CD and cd into C: and access the HDTune 2.55 there (32 bit). By copying   
   one DLL file from System32 on the C: OS, into the HDTune Program   
   Files folder, the HDTune executable can run in a WinPE 10 environment.   
   Doing it this way, is *supposed* to result in the OS not disturbing   
   the benchmark run. The spikes in the trace might be excused as   
   zoned-disk behavior, but I cannot be sure. In other words, I'm   
   still not 100% happy with how these scans look, but I do have one   
   500GB disk which is saintly and using it to test for OS-behavior it   
   would be quite quite clean.   
      
        [Picture]   
      
         https://i.postimg.cc/yxTKRFkb/quiet-HDTune-run.gif   
      
   There is nothing particularly wrong with the second drive. You can   
   see Disk#35 is artificially capped on speed. If it was on the   
   SATA II port, I would think the speed would be faster than that.   
   If it were related to platter encoding or read channel capability,   
   the pattern would have more of a gradual curve. The first SATA drives,   
   left the factory as bridged IDE drives, with a "133MB/sec cap" on   
   speed. That's where the flat line came from on the first generation.   
      
   The yellow seek-dots, are pretty well clustered. I've had   
   drives with anomalous seek, where the drive continued to work   
   reliably. The mostly healthy scans in that example, are   
   not approaching the point where I would be even remotely worried   
   about my JPEGs or vids being corrupted. It would practically take   
   a FireHD or an SSD, to start throwing corruptions like that,   
   and only in the most dire circumstances (approaching end-life   
   on writes-per-location).   
      
   One test I would be doing, is checking whether the   
   hash of the files on the drive, changes on every run,   
   implying a bad SATA cable, or even, a bad SATA electrical   
   interface on one end. I've had one electrical issue on a   
   drive, a 2TB HDD, and I had to retire it before it   
   burned something. There is no way (without a digital scope),   
   to tell the electrical level is outside the allowed common mode.   
      
   The following is a very quick summary, of hashing.   
      
   ******* Some info on hashdeep/md5deep checksumming, just the basics *******   
      
   https://en.wikipedia.org/wiki/Md5deep   
      
      https://github.com/jessek/hashdeep/     # source only   
      
      2014: V4.4   
      
      https://github.com/jessek/hashdeep/releases   
      
           md5deep-4.4.zip   3.62 MB   <=== link is inconvenient to copy, use the   
   link on page   
                                            Need your "good" browser to get the   
   file.   
           Name: md5deep-4.4.zip   
           Size: 3,792,436 bytes (3703 KiB)   
           SHA256: D5E85933E74E5BA6A73F67346BC2E765075D26949C831A42   
   166C92772F67DBC   
      
           You can unpack one of the EXE from the ZIP for usage. hashdeep64.exe   
   on a 64 bit machine perhaps.   
      
     L:\hashdeep64.exe -h                                   # reference to   
   instructions   
      
     cd /d c:\                                              # Set the working   
   dir. Original test tree C:\Downloads   
      
     L:\hashdeep64.exe -j 1 -c md5 -r Downloads > L:\audit.txt   # Generate a   
   filelist with hashes, single threaded   
      
   [continued in next message]   
      
   --- SoupGate-Win32 v1.05   
    * Origin: you cannot sedate... all the things you hate (1:229/2)