HDTune for quick performance tests.
HDDScan (v3.x) and/or Victoria (v4.x, 5.x) for surface access time scan (to reveal bad blocks and slow sectors).

I use something a little later, like a 440BX board, since Socket 7 era was still designed more for 16-bit OSes (or, hybrid like Win 95), and those rely too much on BIOS for my taste. IMO, to really get the health of a HDD, you need software that can speak to the controller directly and expose some of the lower-level functionality.

I use Linux (even recent kernels will boot on a PII without too much fuss), then run "badblocks -svw /dev/sd__" to run through a four-pass (0x55, 0xAA, 0xFF, 0x00) write/read scan. You can use SMART tools before, during, and after on a second console to see if any parameters are changing.

I'm sure there's a more user-friendly way to do this, but IMO, DOS is too far-removed, and Windows is too eager to mess with the disk -- which could have anything on it, if they're not known-trusted drives, and may take forever to start up while trying to read the file system if the drive is in distress.

That said, I admire your dedication. I've bought a lot of "tested / working!" drives that had obviously spent the last fifteen years falling down a Stairmaster, and one floppy drive that needed its capacitors replaced before it would read anything. I always assume the drives I buy will be dead, and it's a pleasant surprise when they're not only functional, but look as though they haven't been used to stop the blades on a blender.

Fagear wrote:

HDTune for quick performance tests.
HDDScan (v3.x) and/or Victoria (v4.x, 5.x) for surface access time scan (to reveal bad blocks and slow sectors).

Thanks

SirNickity wrote:

I use something a little later, like a 440BX board, since Socket 7 era was still designed more for 16-bit OSes (or, hybrid like […]
Show full quote

I use something a little later, like a 440BX board, since Socket 7 era was still designed more for 16-bit OSes (or, hybrid like Win 95), and those rely too much on BIOS for my taste. IMO, to really get the health of a HDD, you need software that can speak to the controller directly and expose some of the lower-level functionality.

I use Linux (even recent kernels will boot on a PII without too much fuss), then run "badblocks -svw /dev/sd__" to run through a four-pass (0x55, 0xAA, 0xFF, 0x00) write/read scan. You can use SMART tools before, during, and after on a second console to see if any parameters are changing.

I'm sure there's a more user-friendly way to do this, but IMO, DOS is too far-removed, and Windows is too eager to mess with the disk -- which could have anything on it, if they're not known-trusted drives, and may take forever to start up while trying to read the file system if the drive is in distress.

That said, I admire your dedication. I've bought a lot of "tested / working!" drives that had obviously spent the last fifteen years falling down a Stairmaster, and one floppy drive that needed its capacitors replaced before it would read anything. I always assume the drives I buy will be dead, and it's a pleasant surprise when they're not only functional, but look as though they haven't been used to stop the blades on a blender.

I don't know why I said socket 7, as I meant socket A - the only suitable build I have for this purpose at the moment.

I will look into Linux. I've never used it, but it does make sense. Thanks

Maybe also look at the manufacturers own diagnostic tools which are often more attuned to specific brand firmware- most of the major brands have DOS / Windows versions eg:

Seagate SeaTools (might cover Maxtor as well)
Western Digital Data LifeGuard
Hitachi Drive Fitness Test (DFT)
Samsung HUTIL
Fujitsu Diagnostic Tool

Either direct from the manufacturers websites, or HDDGURU is a good source for these http://files.hddguru.com/download/Software/

debs3759 wrote:

I will look into Linux. I've never used it, but it does make sense. Thanks

If you've never used it, it can be a little daunting to begin with. But if you find a live CD that works well on your hardware (maybe System Rescue CD?), you can shorten the learning curve pretty substantially. Basically, pop it in, let it boot up, wait until you get a prompt, and type:

badblocks -svw /dev/sda (assuming first hdd -- note this will clobber the contents, so be careful if you have more than one disk installed)

Wait a few minutes to a few hours for it to finish, and power off. You can use ALT+F1 / F2 / F3 to grab additional consoles while it runs, so you can query SMART or what-have-you.

Some software I use:
- Victoria (Windows): I use it with a USB 3.0 "craddle", if I need to use the computer at the same time. Keep in mind that testing may be inaccurate if your CPU load is high (it still will detect bad blocks, but timing may be screwed).
- MHDD (boos/DOS): My first option if the HDD is connected to the mainboard. Being a "boot" tool, it won't be affected by other processes running. It also has some kind of SMART data reader. On the downside, sometimes it needs to put SATA disks on "compatibility mode" (i.e.: no AHCI, no RAID) and (although there is a boot disk that supports it) USB support is... picky.
- gsmartcontrol (Linux, uses smartctl): This is a frontend to smartctl (command line SMART tool) and is present in most Linux distributions (even bootable ones). Those tools can read SMART parameters and run SMART tests (IMHO, SMART tests are not as reliable as MHDD/Victoria tests). USB support is very good, too.

I'd recommend downloading Hiren's Boot CD (it includes MHDD, some "generic" SMART tools and most manufacturer diagnostic tools). If you want to run gsmartcontrol, I'd use System Rescue CD or Knoppix (both are bootable and include that tool). About MHDD, I'd go to hddguru website and download the "Magic" bootable disc, it has better USB support than Hiren's Boot CD (only if you need MHDD+USB).

About HDD block testing...
- A disk that fails SMART tests is a no-go.
- SMART tests don't measure the time that takes reading a block. A block that takes too long to read means a block that probably fail soon. I usually consider unreliable disks that have "orange" or worst blocks (>150ms), but keep in mind (as I said) that if you're using Victoria for Windows and the CPU has high use that may delay read enough to appear as "slow".
- Some SMART parameters are very useful. A high "reallocated" count is a sign that a disk is going to fail; "current pending sector count" may indicate a disk already failing.