Drive question.

[fleggett1]

What do you guys think about a drive that fails a long format (for unknown reasons), but passes all of Scanner's tests, isn't throwing out SMART errors, and comes up fine when running chkdsk?

[Shane]

When you say chkdsk, was that a basic chkdsk or a chkdsk /b (performs bad sector testing)?

I think I'd try - on a different machine - cleaning (using the CLEAN ALL feature of command line DISKPART), reinitialising (disk management) and long formatting (using command line FORMAT) it to see whether the problem is the drive or the original machine. If I didn't have a different machine, I'd still try just in case CLEAN ALL got rid of something screwy in the existing partition structure.

I'd then run disk-filltest (a third party util) or similar on it to see if that shows any issues.

If it passes both, I'd call it good. If it can only be quick-formatted, but passes disk-filltest, I'd still call it okay for anything I didn't care strongly about (because backups, duplications, apathy, whatever). If it fails both, it's RMA time (or just binning it).

YMMV, IMO, etc. Hope this helps!

[Christopher (Drashna)]

Also, the system Event Viewer can give you an indication for why it failed.  In fact, with support tickets, it's one of my first go-tos for troubleshooting weird issues.   If you're seeing a lot of disk errors or the like in the event viewer, it can indicate an issue. 

Also, the burst test in StableBit Scanner can help identify communication issues with the drive. 

[fleggett1]

I've resisted saying this, but I think there's a problem with the Sabrent.  Which, if true, really screws me.

I'm beginning to suspect the Sabrent because I tried long formatting a brand-new 18 TB Exos and it also failed.  I started the process in disk management, made sure that the percentage was iterating, and went to bed.  Got up and nothing was happening and the disk was still showing "raw".  So, at some point, the format failed without even generating an error message.

I'll also periodically wake up to a disk or two having randomly dropped-out of the pool.  I'll reboot the machine and those same disks will magically (re)appear.

I'm currently doing a chkdsk /b on the new Exos after doing a quick format in order to assign it a drive letter (which worked).  It started-out fine, but is now running at less than a snail's pace, with chkdsk reporting that it won't complete for another 125 hours.  Scratch that, now chkdsk is saying 130 hours and it has stubbornly stayed at 5% for the past two hours.

I do have another machine I can try long formats on and will do so, but I'm not sure what that'll prove at this point.  I've also tried consulting Event Viewer, but so much data gets dumped into it that I can't really pinpoint anything (maybe that's just me being an idiot).

I was really, REALLY relying on something like the Sabrent since it seemed to be a Jack-of-all-trades solution to having a bunch a disks without resorting to a server-style case or expensive NAS.  If anyone has any suggestions as to a similar device, I'd love to hear it.

 

[Shane]

Pseudorandom thought, could it be something to do with USB power management? E.g. something going into an idle state while you're AFK thus dropping out the Sabrent momentarily?

Also it looks like Sabrent has a support forum, perhaps you could contact them there? There's apparently a 01-2024 firmware update available for that model that gets linked by Sabrent staff in a thread involving random disconnects, but is not listed in the main Sabrent site (that I can find, anyway).

[fleggett1]

I went into the power settings and disabled USB selective suspend.  I also made sure the system wouldn't turn off any disks.  I just did this, so it'll take a little time to see if it had any results.

I actually had no idea the Sabrent had flashable firmware.  I downloaded and applied the one linked in that thread.  Again, I just did this a few hours ago, so results (if any) may take a day or two to manifest.  Thanks for the link!

However, after flashing, I did try to do another chkdsk /b on the Exos, but it did the same thing and got jammed at 5%.  I'm doing a "clean all" on it now, but if I'm reading things right, that'll take awhile, so I'll leave it overnight.

I'm beginning to think the PSU for the Sabrent might be underpowered, as others in that forum also complained about disk drops, especially when it came to SSDs.  I have all 10 bays populated with various Seagate and WD spinners, which could be causing issues.  If the flash and power settings don't improve things, I'm thinking of ditching it for a 6-bay Terramaster:

https://www.amazon.com/gp/product/B0BZHSK29B/ref=ox_sc_act_title_1?smid=A307CH216CTGMP&psc=1

I don't like the fact that the Terra seems to use old-school drive sleds, but I'll gladly accept that hassle if it means I can get the pool back to 100% (I can settle for only six disks in the meantime).  I might even take apart the Sabrent to see if the PSU can be upgraded (probably not, but it's worth looking into).

More to come!

[fleggett1]

I think I'm gonna give up on having a pool.  Maybe a computer.  I was trying to clear that 18 TB Exos out using diskpart.  This Exos has the label of ST1800NM.  Another 20 GB Exos that I had bought a few months ago has the label ST2000NM.

I was tired, bleary-eyed, more than a little frustrated with all these problems, wasn't thinking 100% straight, and selected the ST2000NM in diskpart, and cleaned it.

Problem is this drive had gigabytes of data that was critical to the pool.  GIGABYTES.  I still can't believe I made such a simple, rookie, and yet devastating mistake.

My God.

I don't know if any of the file table can be salvaged, as I just did a simple "clean" and not a "clean all", but I've got a recovery request into Rossmann Repair Group to see if they can do anything with it.  I know there are some tools in the wild that I could probably run myself on the drive, but I don't trust myself to do much of anything atm.

I should never have thrown money at an AM5 system.  I also probably should've stayed away from the Sabrent and anything like it.  Instead, I should've done what any sane person would've done and assembled a proven AM4 or Intel platform in a full tower and attached the drives directly to the motherboard.  Yeah, the cable management would've been a nightmare, but literally anything would be better than this.  My goal of staving-off obsolescence as much as possible has instead kicked me in the teeth while I was already lying prone in a ditch.

If, by some miracle, Rossmann is able to recover the data, I'm going to take a long and hard look at my PC building strategy.  Hell, maybe I'll throw money at a prebuilt or one of those cute HDMI-enabled NUCs that're all the rage.  I just know that I'm exhausted and am done with all of this, at least for the time being.

[Shane]

It's an awful feeling to have when it happens (been there). Best wishes, I hope you get good news back from the recovery request.

[fleggett1]

I haven't heard from Rossmann yet.  If I don't get a reply by Wednesday, I'll voice call them.

I did order the 6-bay Terramaster, which should be here tomorrow.  If/when I'm satisfied it's running as advertised, I'll rebuild the pool and copy over what I can.  Assuming everything goes smoothly (har har), I'll reinstall everything and restore from config backups, which I've been holding-off on doing until I can figure things out.

I really, REALLY hope the Terramaster "just works".  Since it only accepts six drives, the power demand should be a lot less than the Sabrent.  I'd still like to disassemble the Sabrent to see what sort of PSU it's using.  I'm 99.9% sure it's something custom, but you never know until you actually eyeball things.

I'll report back in a few days.  Wish me luck.

[fleggett1]

Rossmann said they couldn't do anything with the drive, which REALLY surprised me, so they referred me to Gillware Data Recovery.  I need to talk to them first before sending the drive in to make sure the attempt will be a reasonable sum, which I'll try to do tomorrow.  I guess the moral of the story is not to futz around with diskpart unless you know for damn sure you know what you're doing.

I received the Terramaster.  It looks like a typical grey internal drive cage, just a little wider and without any mounting points.  It uses screw-in drive sleds, unlike the Sabrent, which is a PITA, but c'est la vie.  However, it also came with some paper-like inserts that go in-between the drives and the bottom of the sleds, which I've never seen.  The boxed instructions don't say why they're required and the online product guide doesn't even mention them, so I can only guess their purpose (electrical insulators?).  Bizarre.  It also doesn't have an internal PSU as such, but came with a laptop-style 12V barrel plug brick power converter.  Unlike the Sabrent, the drives sit vertically, which makes me a bit uncomfortable (I might lay it on its side).  There is only one USB-C connector.

There has been some talk on the Terramaster forums about people needing a powered USB hub with this unit.  I don't know why this would be required when directly attached to the USB-C port on a motherboard, but maybe that's been my problem all along.  If so, I'll consider slitting my wrists later.

I'm in the process of long-formatting three 18 TB Seagate Exos simultaneously.  If that passes, I'll continue trying to rebuild the pool.  This is my last attempt at doing so.  If the formats fail, I'm giving-up, at least for awhile.  Maybe in the far-flung future I'll get a proper tower with a 10,000 watt PSU and stuff it with drives, but I really, REALLY hope the Terramaster comes through.  I have confirmed that Windows recognizes all six drives, but the Sabrent did the same until I started encountering those drive dropouts, so longevity will determine the winner.

More to come!

[fleggett1]

The drives long-formatted just fine.  That's one hoop jumped.  Now, I can start rebuilding things, but I need some pointers.

Three of the drives in the Terra are part of the (old) pool and is being recognized as such.  I want to create a new pool using the three Exos as the base, but I don't want DP to have to contend with two pools going simultaneously.  I know the software is designed to support multiple pools, but I want to keep things as simple as possible.  Plus, in the end, I only want the one pool.  Once all the data is moved to the Exos, I'll be long-formatting the old drives to ensure they're still good.

How can I eliminate the previous pool without affecting the files on the three drives?  Do I just remove them one-by-one, presumably in archive mode, until they've all been processed?  And is there anything else I should do afterwards, like taking ownership of the old pool files and/or resetting permissions?

Thanks in advance!

[Shane]

If you're wanting to keep things as simple as possible and you're going to long-format the old drives anyway, then I'd suggest trusting DrivePool to be able to handle the job?

1. create the new pool
2. copy your content from the old pool to the new pool
3. remove the old pool
   • (quickest way #1: stop the service, rename the poolpart.* folder on each to oldpart.*, start the service, remove the "missing" drives from the pool until all gone)
   • (quickest way #2: eject the old drives, remove the "missing drives" until all gone, format the old drives on another machine)

This has the side benefits of #A, not having to worry about manually fiddling with pool parts and/or resetting permissions, as unless you're deliberately copying across the permissions from the old pool your content should inherit them from the new pool, #B, you can optionally run lennert's treecomp or other comparison tool of choice after step 2 to ensure it got everything, bit-for-bit if desired, and #C, give the new Terra a good workout now rather than later. And at the end you'll have one pool.

P.S. If you've got the drive capacity now, consider turning on real-time x2 duplication for the new pool. YMMV but even though I've got nightly backups, knowing that if any drive in my pools just decides to up and die out of the blue I still won't lose even one day's work from it gives me extra peace of mind.