That reseller warranty and recertification process is most likely, Step 1: buy a laser engraver Step 2: claim you recertify drives and brand them as such Step 3: offer a warranty, so people feel safe. Tommy Boy said it best.
He said that there’s proof of recertification when he checked the drives. So their claims of recertification was correct, even for the drives that had listings that didn’t say they were recertified….were actually recertified. Buyers will notice and you won’t be selling them for long. It ain’t that easy anymore.
@@sultanofsick yeah you May be right but like he said, reputable sellers are selling them and buyers are satisfied. So it doesn’t seem like sellers are actively trying to scam buyers. I think both buyer and seller know the deal, both know the trade-off that’s explained really well in this video.
Generally SSDs are much better, there's no moving parts, so the only thing to likely fail is a dried up capacitor, but that could happen to anything and the NAND itself can fail, but unless you're looking at QLC, most drives are rated at roughly 50-100TB written per 100GB, so aside from defects or farming crypto it would be pretty hard to kill the NAND itself. The real problem with used ssds, particularly old ones, is going to be the lack of dram cache and they often aren't all that much cheaper than just buying a new one, at best you're looking at double the storage for the same price but on an older/slower drive for example I've seen the WD SN750 500GB go for $60 new and the SN550 1TB $60 used, that's not bad, but I've seen much worse like older sata drives lacking any dram going for the same price as those two newer WD drives.
If it is SSDs used to plot chia they will be thrashed. While these HHDs used to store the coins didn't do a lot of read/write the SSDs used to actually plot will have tons.
I personally wouldn't trust used SSDs for anything other than content you've got backups of or don't mind losing. As with any solid-state storage devices, they can just die without warning and you don't have the acoustic reporting to tell you if something is wrong.
If it's SATA, I've been getting the Samsung 850 Pro from 2016 or so (10 year warranty, so still valid!) and SM863, also from the same era. Those 2-bit MLC V-NAND chips from that time have an amazing endurance, worsening as they added more layers. The SM863 is enterprise so includes power loss protection capacitors. As for reliability, all SSDs seem to have some troubling avenues for losing lots of data easily. I stick to just having the OS and applications on it. User files, media, etc. on an HDD. At least the enterprise SSDs probably have more robustness (underprovisioned more for more spare sectors, more solid ECC).
I see no point in buying used SSDs.. you can get new off brands dirt cheap; which do fairly well, and you don't know how much data was written either, unless specified. SSD do fail once they reach their limit. Plus they are hard to detect when they fail, can die any time; had some good brands die on me before.
Just an inconsistency note: the table at 5:38 indicates that the WD drives indicated all 5 SMART fields of concern, but Alex says right after that it was Seagate who covered all 5 SMART fields of concern, not WD. The screengrab shown when he says this seems to agree with Alex rather than the editor on this one. Edit: lots to keep track of in this video, quite a good job overall
That serverpartdeals seller has always been pretty good in my experience. I've gotten a few 8tb wd enterprise drives from them, and they're always super responsive
@@theironangel767 SAS drives are meant for enterprise use so they're going to be built better than even prosumer SATA drives for NAS use. Frankly, this video really undersells the longterm reliability of used enterprise hardware. However, SAS requires a bit more knowledge in order to get working in any sort of setup.
Yeah, I bought 9 of the Exos X18 16GB drives recently. Had 2 failures they replaced, and replacements were good. Both bad ones had failed with uncorrectable sectors during initial testing.
@@randomman057 Sure does, 2nd hand enterprise is the way to go for a cheap build out. You can get some great deals if you buy irregular format drives and know how to reformat them back to 512b - as long as the drive supports changing their allocation, some are locked, especially SSDs. I got 10x6TB 520-byte drives a few months back for $45 per drive, low writes only a few thousand POH and from late 2018/2019
Chia is very lightweight on storage drives. So if the eBay listing specifically states it was dedicated to farm chia it is most likely better than the majority of other used drives out there.
Crypto bros clearly don’t understand how an economy functions, so I highly doubt they know a single fucking thing about drive loading and wear. Besides Chia is currently worth $28.78USD, so not only is your pretention hilarious, it also highlights how fucking utterly clueless you cryptobros are.
I've gotten some amazing deals on used hard drives. Got like 5-6 Firecuda drives that I'm pretty positive weren't even used. Payed like $10-$20 a piece for 1-2 TB drives.
Yeah I was about to get new drives for my nas but I just got a deal for 1tb drives at 11€ a disk shipped. Got 7. So it definitely can be great. I have had bad experiences with used drives so not always great. 2 years ago bought 9 500gb drives, on arrival only 6 were working, got a partial refund (ebay is always great with that, even if seller is an ass, ebay will step in and find a solution) and now 2 of the 6 that were left are dying. Bought another drive (to replace one of the two that was dying) and the thing literally came in a letter, it works and tests fine but I don't trust it now.
@@imnota it can be hit or miss, that's for sure especially when you find them for sale really cheap. When they're certified refurbished and closer to the market price it's a safer bet but when you find a really good deal I'd basically always consider it a gamble.
@Jake Out with the old in with the new. Everyone wants that high performance of SSD drives, brings new life to even old PCs, like going from HD DVD to Bluray 📀
If you do a follow up, I encourage you to cover usage of tools like badblocks and hard disk sentinel's reinitialize test for burn-in testing drives. In the dozens of new and used drives I have had, it has been very reliable at catching early failures before the ebay money back guarantee/newegg return window closes. None of them that have passed this test have failed before they were re-sold years later, personally. Cheers!
HDS is an amazing utility although it does take a long time to learn all the nuances of its different test methods, what types of errors they expose and what they don't. Essential if you care about your data storage or spend time fixing computers a lot.
@@mubok3743 I wonder how many sellers are experienced to wipe out the SMART data, and reset them back to factory-new-like condition. Because its quite easy to fake a new hard disk if you reset its SMART counters.
Used to work at a little computer repair shop. We had all kinds of hdds from SSD upgrades that people didn't want. One time a youngster came in with $3. We have him a box and said he could take 10tb he was so happy
I bought a 4TB WD Gold HDD on eBay about 4 years ago for £60. It's still working perfectly. I risked it because it was reasonably cheap and as it's a WD Gold, it's intended for datacentre use so it's designed to last long (MTBF of 1.2 million hours!) and is less susceptible to mishandling. Buying a used hard drive isn't always a bad thing. Edit: I just checked how long my drive has been running for and it's currently at 54,480 hours, SMART status is all good.
For Chia, just writing the final plots to the drives doesn't wear them too greatly, but if they were used as the temp drive to make the plots, then the write amount goes up exponentially.
@@quantum8522 it's not as bad as you think it is. I didn't have the budget for high capacity SSDs coz they're expensive here, but I used multiple HDDs each with its own queue up to what my RAM allowed and they did okey. It wasn't economically feasible when people where farming solo and trying to beat the growing network space, not to mention the declining chia price.
I lucked out on some WD Reds that had chia plots but the price was 1/3rd of the original and they had like 300 days total work. For me it just passed the infant mortality stage and it was already pretested lol
Every "factory refurbished" Seagate drive I have bought from ebay had its SMART stats, including power-on hours, reset. Buying a used anything is always buyer beware. But I have actually received several brand new DOA drives while all of my ebay finds are still working after years of 24/7 operation, so your mileage will obviously vary. RAID6 or z2 is a must if you care at all about the data regardless of whether the drives are new or used....
I have a 30gb HDD from way back in 2008, that still works relatively well. It's used right now as a storage for some games like XCOM and other, older, titles. It's being read and written to somewhat infrequently, as it's not a boot drive for anything, but still, kind of amazing that it still works.
@@leprechaunbutreallyjustamidget That HDD was the first one i got as a present from my dad. Still kinda keep it as a memory, pun intended, along with other things i got from him. I do have a normal 1tb M.2 drive on the motherboard (got it on the cheap, as the company that sells PC parts went out of the business, and were now offloading the leftover parts) and a 2TB additional storage HDD, so i'm not putting valuable files on that ancient hard drive. Just keep it around for the memory's sake, as that little present opened up a whole world of PC building for me.
I grabbed a whole bunch of Seagate x16 drives that were ex-chia mining for all around $200 a pop, bargain. Very glad you guys did this video, helps educate what to look for. Fortunately my seller did amazing packaging, but another issue here in Aus is drive on Amazon shipping from US basically loose in a box. Really bad.
Even on the fast internet I have, it can take over an hour to download larger games. An inexpensive HDD is a pretty decent solution to keep games stored locally if you are not currently playing them and want to keep the ones you are playing on your SSD.
A Steam Cache is useful too. It's like a separate little server that you download games locally from instead. Pretty neat, especially if you play a lot of AAA games.
only one hour to download larger games? That's...insanely fast already. Took me 11 hours to download Cyberpunk for example lmao. But yes, having a large HDD to always have all your games ready is a must. Using a used one too
@@reptilez13 Steam Cache is still slow as shit tbh. I run a pc cafe and we used to have a steam cache running, but even at 1gb/s (the limit you will hit on most affordable multi-port network switches), very large games would take 30 minutes or longer to transfer over and most people like to try a handful of games before settling on something. We've since migrated to using 1TB SSD along side a 14TB HDD internally for each station. The SSD and HDD are tiered, so the most played games basically stay on the ssd and even if you choose something less popular, at most you'll have to wait through a HDD loading screen once, before the game moves over to the SSD automatically while you play. Waiting a few seconds more for a loading screen is much better than waiting half an hour for an entire game to transfer over. Even manually moving games from HDD to SSD is faster than the 1gb/s bottleneck you hit with a network. People consider hard drives to be super slow, but high capacity hard drives can easily hit 200MB/s or higher when they aren't running an OS and the workload is mainly sequential. Having an SSD cache and 10Gb network infrastructure would have also been a solution, but even that wouldn't have been as fast as our current setup and would have cost 10x as much. We would be better off putting 10TB SSDs in each station and calling it a day.
Adding to the anecdotes: From my work experience enterprise drives rated for 5 years usually hold up at least 50-60k hours and I'm currently emptying some drives on a personal server that have 72k hours and still run just fine(the server just got superseded). BUT some hard drives just die in the span of a year or two. It really just depends on luck more than anything.
I got a Toshiba 16TB Enterprise helium-filled drive through Ebay and it had just a couple hundred hours shy of one year's worth of hours and only something like 20 on/off cycles. It was perfect! (still is). Quiet and super-fast. I liked it so much I ordered another one and never got it to be recognized by the computer. I wrote them and they wrote back and asked me to send it back... super easy. I got reimbursed and ordered another form the same people and it went in just fine and works as well as the first, but this one had about a year's worth of hours and 24 power-cycles. It is as quiet as the first one. All three were sent pretty well packaged with sealed ESB bags and foam around them. I also got some SSD's (4TB Samsungs) for about $60 off regular price, but 'open box and 'tested'. They came in envelopes, but packed with cardboard, ESD bag sealed, no label or original box as they must be kept for records. One power cycle and 67GB written and erased, otherwise, just like new ones. And since these things might last for decades, I figure as long as they are in good shape when I get them, they'll probably stay that way.
The key thing here IMO is the power cycles. The fewer the better, and most of these resold enterprise drives haven't even gone through 1/10th of their MTBF rate. Chances are the drive will outlast your usecase and you should be RAIDing them anyways. I six drives from GoHardDrive like 5 years ago and loaded them up with ISOs in an array. I've since sold those drives and they are still chugging away in my friend's server without issue.
Same case here got couple of WD Reds for cheap filled with plot files. Power on time 300 days and 30 power cycles still work till this day. I have read that chia storage drives are actually the best second hand drives you can get because they don't see much work after the initial writing
An important warning you missed, particularly in relation to shucking drives - if you plan on using any ZFS based OS (Xigmanas, freenas etc) DO NOT use any SMR drives, which are commonly used in those portable hard drives that people shuck, which is why they're cheaper in the first place.
@@snowhusk SMR stands for "Shingled Magnetic Recording". As the name implies tracks on SMR overlap like shingles. Initial writing on SMR drives is pretty reasonable, but as soon as you delete files things start to get unbearably slow since it has to move any tracks that are overlapping on the file you're moving/deleting first. I've started buying them for long term backup/storage in place of WD green drives because the price:byte ratio is exceptionally good, but again I wouldn't use them for much else than files you don't intend on *ever* moving around or deleting.
anecdote time: I bought two used hitachi 2tb drives with ~10,000 hours on them like 10 years ago and still am using them. Now they have ~92,000 hours. Idk if i was being stupid, but I actually bought them with the mentality that if they were still working after being used for a while, then they probably weren't about to fail.
Did the same with a decertified 8tb easystore. Used crystal disk and it said it had like less than an hour of use and a handful of power ons. All for less than $100! Great and cheap, perfectly fine for mass storage!!
Oh those drives usually have their SMART data wiped. 1 hours powered on is a red flag. That said, there's nothing wrong with throwing them in a raidz2 array. Even with redundancy they're much cheaper than new drives
@@NavinF oh wow I had no clue haha. Good to know!! Thankfully I’m just using it to store footage and photos from projects/events I’ve finished editing, so I’m hoping it holds up over time! Thanks for the heads up!!
@@syarifairlangga4608 There is no "SMART sensor." The drive keeps its SMART data on a normally inaccessible area on the platters and it's not impossible to clear it
IMHO, if you have data to archive (not back up), let me repeat, archive, then WORM storage is the ONLY solution. Enterprise often uses LTO, but I have found that DL BD-R is great for the average consumer. I have several TBs of data archived to DL BD-R, and I fully anticipate that each disc will remain readable for at least 20 years. Of course, burning, verifying, cataloging, storing, and reading is tedious and time consuming, but too many people treat data as an afterthought and that sort of carelessness is why data get corrupted and lost. People only care about their data once it goes poof. Learn to care about your data before it goes poof so you'll invest the time and money to ensure it is secure. Once you've done that, failing hdds will be the least of your worries.
I got curious when Alex started talking about power on hours, and the expected lifetime of drives being 45,000 hours, and setting a ceiling of 25,000 hours for purchasing used drive, so I opening up crystal disk info and took a look at the power on hours of my drives. I was absolutely floored to see oldest drive in my system, the one that I use for the majority of non critical software installs, has a whopping 70,000 hours.
I bought 8 x 3TB HGST Sata drives via Ebay in 2016 and set up my home server. Those all had about 20K hours each and were running 24x7 from 2017 until spring 2022 when I changed my server setup, no issues, no failures. New setup is 4x 10TB HGST SAS drives, ebay for $80 each, 12-14k hours each, running flawlessly since I set it up about 6 months ago. For the price, I keep, depending on the box, hot spares or a cold spare handy just in case. Far more effective from a cost standpoint even if you happen to have a drive die.
Not even used drives, I've sometimes found brand new ones dirt cheap. I've seen WD enterprise grade drives for less than £10. Sure they're only a couple of hundred gigs in some cases. It's also worth noting that you can't necessarily trust shucking some external hard drives, especially the ones meant to be accessed remotely, as they can be locked to only work with the controller boards attached to the cases. Another thing worth noting is that you should only ever use the numbers provided by programs such as CrystalDiskInfo as a rough guide. I've seen conflicting reports between different programs, some drives don't even report some numbers, illogical Power On Count vs Power On Hours, POC and POH numbers even for brand new sealed drives, and it's possible to tamper with the firmware to report less than is the actual case.
Like Alex said in the video: Ebay protection is very good for consumers. Make sure you read, and fully understand the policy before you buy anything. that way if you end up in a situation where what you receive doesn't match what you were told you were supposed to receive, you can tell the seller "your listing says no return, but ebay's policies say this, so i'll be returning this, send a shipping label" and their options are do it, or you go through ebay and get your money back regardless.
Great video. I recently bought 4 4TB WD Reds for my NAS from eBay (taking it to the max size, stinkin' 4TB limit), supposedly former pulls from a NAS, and while I didn't check beyond a SMART Short and Conveyance test, the attributes were fine, with no ATA errors in the log. Paid $30 a drive when they're easily quadruple that cost for a new one. As long as you're running in a RAID mode that can survive 1 or 2 drive failures, you'll be fine. For full on failures like yours Alex, look into RT Tools's R-Studio, amazing recovery tool.
This exactly. As long as you aren't buying decade old drives and purchase them from reputable sellers you probably won't have an issue. Used HDDs are perfect RAID 5 or 6 configurations. It's definitely cheaper when the cost of replacing a single drive is less than half the cost of new. Buying used in bulk leads to crazy deals. I've regularly seen 10-20 or more 1TB drives selling for under $10 per drive, sometimes as low as $4. Hard to go wrong with savings like that.
So, I purchased over 60 used 4TB hard drives from eBay for Chia farming (no it hasn't been profitable). It's been about a year of continuous use. Very few have failed! It's hard to tell which have truly failed and which has lost connection due to USB dock issues. Install is under a bed so it is not easy to dissect. I lost connection on a number of them in the early days but for about 9 months now, farm size has been consistent. Majority of them have 40-60K hours on them when I got them. Some are WDs all 7200, some are Hitachi 7200, some 5400, some are Seagate 7200, some 5400. For unimportant storage, it is definitely worth it the reduced cost/TB and reduced carbon footprint of not having new drives manufactured. Though, it does make you wonder about the people unable to buy the drives I bought, if they ended up buying new negating this benefit? A note, I also have a huge Distributed Computing farm (Linus Tech Tips team) and 1 of the drives in my PCs (labelled Haswell#11 in BOINC) is a WD2502ABYS Enterprise grade drive with 86.5K hours on it. SMART report is perfect! Some drives are engineered really well! I say that power cycle count is very important to judge drive wear. Temperature cycles causes wear, whereas if it's only had like 25 power cycles, you know it's been at a consistent temperature through most of its life. Almost all my drives do have a low power cycle count.
This video made me check, I still have a usable hard drive with 45000+ h of power-on time. its a slow as shit 5400 rpm that i have kept migrating between systems for years and that has seen very little to none usage in the last couple of years (hence the fact that it still works at all), but still, its running in my system right now and I recently used it as a buffer to move data between ssds without issue.
@@dadodi I'm assuming it's making a quick back up of stuff when moving files to another drive. So so you're transferring from SSD 1 to SSD 2. The aforementioned hard drive would serve as the back up buffer so there are multiple points of failure. It's a good and simple way to do things. However, with this method, it will take a little more time.
I have a few sitting in the 40,000's hours. The 'champion' however is a 5400 RPM drive in a work PC that I built which is nearing 10 years of spin up time. All its SMART parameters are OK but I'm hesitant to do any testing on it all. One day it will be layed to rest but not today. I think the very lightweight platters in these slow 2.5" drives gives them an edge into living a long life!
@@indoctrahol that too, but i was mainly using it as a way to temporarily store the data because having the 2 ssds in the system at the same time wasn't an option because of limited m.2 slots (and no other drive having enough free space)
Note on shucking, do not buy the Western Digital Elements line for that. The logic board on those drives is USB, there's no SATA port on it; if you want to use that inside a computer with a USB 3.0 header to type A or C or whatever port and hook it up internally over USB it'll work (in fact if you want to do that because you ran out of SATA ports, maybe _do_ do that), but not in a slot loading bay that hooks directly up into a SATA connector that's permanently mounted like in a standalone NAS enclosure. I did not learn this from experience in frustration trying to do that, but I was considering shucking one myself thinking it was a replaceable drive inside an enclosure like the My Passport line, and learned about its contents while looking for instructions on how to open it without destroying any and all plastic clips holding it together.
Not true only some are but you should be researching anyways before you buy because same model could have different model hdds some faster than others etc some with helium etc
I bought a "new" hdd on ebay awhile ago. I checked the crystal disk info for it and it had been used for like 2 years. You really got to be careful with buying hdds.
@@gorkskoal9315 Unless you mean toshiba ones. It seems they have a tendency to early failure, but just reliable as anything else later (and better then seagate, so there is that).
Hi Alex, I just checked 2x WD disks though via USB as it's external. Linux Mint -> GSmartControl (have to install but it's in default repo) -> Attributes Tab shows no LBA, but click on the next "Statistics" Tab and there it is: "Logical Sectors Written" and "Logical Sectors Read", both with Values. So it seems WD disks are collecting LBA Written/Read but shows them somewhere else.
If you buy an used hdd, let's say from Sweden for example, how long does it take until it catches a torsk smell on atomic level? Does it require to be wrapped inside an ylletröja to make it operational? I'm from Finland btw.
I got a "refurbished" 12TB drive from Amazon the last week for $125. But according to the SMART data it was brand new and that was the first power up. Not sure what's up with that. The drive came in a static-proof bag inside a barely padded envelope. When the delivery driver dropped it on the front porch, I heard it hit the cement from my home office next to the front door. Been running without issues for about a week now. I basically copied all my old drives onto this one drive and I still have the old drives, in case anything goes sour.
@@petedavis7970 Beware that for some models the Smart can be reset, even if it had bad sectors. I did this to WD drives. There are utilities but hard to find and are paid. On old hdd, I would advise to write some data or just read with Hdd Scan and keep it offline for a few months and read again and compare the timings for sectors. If it gets high timings, like 70ms, 120ms on many sectors, it means surface is degrading and got weak and if you write data there and don't overwrite it in a year, it may be unable to read it back (weak sectors). Note that even new hdds can have this issue, weak surface, so new is not always better. There is a utility DiskRefresh, which reads and writes back each sector, to remagnetize the area. You can use that from time to time, like 6months, to prevent silent corruption (bit rot).
As long as one is able to find sellers who regularly deal in white label hard drives, with their own warranty and customer service to back it up; they can be a relatively safe option for maximizing storage capacity for those who wish to keep their data local.
Amazing how often the warranty coincides in length with eBays money back policy. Literally no real downside to offering returns for faulty drives for 30 days when eBay basically does for you, but without the return if they say so.
I mainly use decommissioned hard drives for cold storage backups or for non-critical use, like a Plex server. When I get it, I run a thorough test of the drive, then copy a bunch of data using a USB 3.0 drive dock. When full, I put them in a storage case and store them on the shelf. Can't accumulate power-on hours if they are offline, nor can they be encrypted by hackers. I also use LTO tapes, USB thumb drives and SSD drives so I have multiple copies of important data on different media for cold storage backups. I recently picked up a 20TB rebranded Seagate drive for $199. Tested fine, copied a little over 19TB's of data to it and now it's in a storage case on my shelf. Multiple backups on different media, stored onsite and offsite is key.
Could you guys do more videos to explain certain aspects of tech? Like the different kinds of ssds and what use cases they are best for? Or the different kinds of panels in Tvs and monitors, i.e. tn vs ips vs oled vs quantum vs whatever else? I know I can google all this eventually but I would never remember it all anyway and it would be sick to have a comprehensive video that I could reference about like each category of tech. You guys do pretty good work covering what you cover I just wish you would stop making sub 15 minute videos. Idk if you tend to do this because it is faster for production or the TH-cam algorithm tells you this is the best way to make money but I follow this channel to learn. So don't rush the video lengths, you don't need to, I'll eventually watch it all!
I've built a bunch of sketchy servers with raid 6/5 arrays using old drives and *knocks on wood* they've yet to let me down after over a year of 24/7 service. IF you use a raid array of older drives, one helpful bit of advice is to always have 1-2 extra drives set aside for each machine to quickly replace one of the active ones if it happens to die. Raid is not a backup and as long as you have a contingency plan, I see having a bunch of older drives that I feel ok discarding at the first sign of failure to be a very safe bet.
Great information as always. The only thing I would add is that Amazon sells the WD enterprise drives refurbished and they do accept returns within a reasonable amount of time. I have had no issues with the “Renewed” drives as they label them.
Full time chia farmer here. Chia doesnt really hurt the drive. Once its filled, it might get a read once every few minutes. Theres higher risk storing the hdd on a desk since you camn drop/ bump it/ break it easily.
The best way to get _stupid_ amounts of cheap storage on eBay is to get old SAS drives. They are often dumped to eBay in bulk _way_ before they are actually used up, they're usually 10k to 15k ROM drives, they're usually _ultra_ reliable, and they are sold for so cheap because they are usually formatted in 520 bit sector sizes and require a SAS controller to use. So, what this means is that you will need a special controller card for them and then you will need to do a block reformat to get them back to a 512 bit goodness that most home PC's can deal with. The controller cards can be had for pretty cheap on eBay, but you do want to make sure that you get a model that can have the firmware flashed to be better for non-enterprise use. And block reformatting is super easy once you figure it out and you can queue up as many as you can plug in before going to bed or work or whatever. And from what I've read, what happens a lot in enterprise data centers is that drives are basically _never_ run until they actually start hitting their r/w limits and instead often just get thrown out and replaced whenever they get full or close to full, just because hard drive failures are so catastrophically bad in enterprise settings that it's just not worth even the miniscule risk of trying to squeeze extra use out of them plus it's just a pain in the ass to try and keep track of which drives are oldest and need to be replaced, so when they increase their storage they just replace their existing drives with higher-capacity drives rather than add more drives on top of existing hardware. So they dump these drives that are already rated for super-high reliability when they still have about 95% of their useful life left. Definitely worth looking into if you're like me and you have a lot of storage needs and/ or want to have a private cloud storage server and you're willing to tinker. P.S. it's also worth looking at old SAS SSD's to use as high speed cache.
6:10 Bad backplanes can and do cause corruption. I've experienced it myself more than once, stuffing known good drives into cheap 30-bay NAS cases and getting hundreds of errors in literally minutes. Data was irrecoverably lost, because ZFS and similar file systems will try to correct the errors by overwriting them with data from the mirror, which if it doesn't itself come back corrupted, may write things out corrupted. IIRC Linus himself experienced this exact issue a few years ago with his "storinator", it's a bit weird to just dismiss the problem like that.
11:14 Warning on external drives meant to be used for backup. A LOT of them (likely a majority by now) WILL BE SHINGLED DRIVES. I've had 4 external 8 TB drives - 2 were Seagate and turned out to be ARCHIVE models, 2 were WD and turned out to not be SMR (but they WERE advertised as "GAMING" externals, not something like the MyBook backup drives).
@@alexturnbackthearmy1907 Why do you ASSUME that singled drives didn't get mounted in a case or rack? Both of my Archives have been in racks for YEARS now (it was cheaper when I got them to buy the external, then "shuck" it and toss the parts I didn't need to use).
@@alexturnbackthearmy1907 The only way for drives to get "thrown and shaken around" would be if they're NOT in some sort of a case or are in portable cases.
@4:30 I'm pretty sure this is how Toshiba would normally ship items out considering their "quality control" and life span... (Same with HGST). Terrible spinning drives IMO. That goes for both Enterprise and Home Use.
I wanna build or buy my own NAS. But the single most probably main reason I'm hesitant to do so and still using cloud storage is the AI capabilities of Google photos. The main reason I opted for cloud storage is that whenever I take a picture it's set to automatically be backed up to Google photos and I know I can find alternatives to initiate automatic backups. But the search feature in Google photos is irreplaceable to me. I can search and find almost anything in my photo collection. I can search for specific people, places etc. Is there a way to search and sort everything in my NAS like I do in Google Photos.
Should be noted that when using external drives they will be slower and have a firmware optimized "safer" operation, as there usually "B" or "C" grade drives from the factory as not all drives come out the same from the assembly line. I have to do data recovery on one or two a year because clients use them as there main work drive, they usually just last about over a year when actually used every day
Gee! [in a tone of puzzlement] Clients using external drives as their main work drives is a dicey thing to do, given that they could be misplaced, dropped, or stolen in the workplace. All that _read/write_ on an external drive? They weren't built for that.
Those external WD drives are great for the price. I made an 8TB backup NAS using one and a Raspberry Pi 4. Sadly, getting your hands on a Pi is near impossible these days but you could easily use a Radxa Rock 4, whilst a bit more expensive(£60 instead of £40) it'll do the job just as well.
@LTT - just a fyi, all the "Storage Based" Mining coins are worthless on HDD's. Potentially you could use ZFS and keep you ARC on SSD's to overcome the issues with this and flush the data to HDD's. But, you would still need relatively few HDD's, and a excessive amount of SSD's to make the bulk of it. As it requires fast, no-latency storage to function. (In that way, its significantly worse then GPU mining) So while its remotely possible some one made a huge array of these drives to try and overcome this issue with a massive CPU and a Abundance of HDD's. The more likely scenario is, small to medium businesses (and former small-scale datacenters) are dumping their old drives and matched spares onto the consumer market, after either replacing them or the company preparing to liquidate. -- Here in Australia, we see this a lot. Ever 2-4 years, there is a massive influx of HDD's on Ebay and Gumtree. Most of these are smaller businesses refreshing there storage with newer sku's. As they cannot get replacement spares from the same series. In IT as im sure your aware, Matching your drives is super important to provide predictable performance and reliability. Thus having hot and cold spares from the same SKU lineup is a absolute must. | Given the lifetime of such drives being around 3-5 year max, its often after 2 years you see 2yr power-on drives dumped and at 4 years a mixed 2yr power-on and their spares dumped on the market. As to why the consumer market, and not liquidation outlets/auctions. Thats the simplest question. Profit. Auctioning HDD's in Lots, results in the overall unit price being low. Also, without consumer protections, less individuals (Datahoarders/HomeServer) and businesses are interested in investing. So you see these go for minimal amounts. | So unit prices on Ebay means top-dollar for your used (and unused spare's). Perfect example of this is in 2019 i acquired 32, 8TB Drives for $17/aud and 24, 12TB drives for $8/aud from two separate Grays Auctions. | from this 30% of the drives had never been powered on. | my savings? around $6000/aud, compared to retail.
You should do a video on used enterprise SSDs. I think they might be worth it because the good ones use MLC flash which is superior to TLC and QLC found on most consumer drives. They also aren't as suceptible to physical damage during shipping.
For enterprise drives the ones with capacities like 400 and 800 GB instead of 480/512 and 960/1TB are better as they have much better wear leveling. Even better if you can get SLC ones.
If you have tons of money, sure... If not, I recommend some sort of RAID redundancy setup and monitor them for read and writes... ZFS is very critical and will let you know when they start to notice errors, at that point you swap it. The real question is, how paranoid are you, and how likely are you to lose data, I went with 2 NAS boxes to satisfy my paranoia after my 1st NAS box which was a RAID 6 at the time, had a harddrive failure and a software issue that kicked out 2 drives, leaving me with no redundancy during rebuild.
Bought a wd green ssd from a car boot sale for £5, turned out to be dead. But I checked the warranty status and it was still under warranty so I sent it in for RMA and got a new one in return.
2:14 Storj does not "mine" on a hard drive. It is a BACKUP setup, similar to a low-budget BACKBLAZE, and those drives often see NO activity for months once they fill up. Tends to cause even LESS wear than something like Chia.
I think testing AliExpress HDD's or even better SSD's would be interesting. I've seen many high storage SSD's for dirt cheap (2TB M.2 for 52€) and im wandering if they're any good.
Imo: New SSD, keep personal stuff on that + backup on external, cheap high capacity ebay drives for games storage and just backup your screenshots if you care about those, can't go wrong! At least it isn't a quantum bigfoot drive..
I've suffered the drives sent wrapped in the thinnest of bubble wrap from Ebay. They are the worst, but as you say at least your protected. Selling any kind of IT kit on Ebay, I always overpack to protect the item and take a picture before sending. Apart from the odd nutcase buyer, never hadxan issue. Really enjoyed the video.
Got shipped a RMA return discount drive from probably the most trustworthy retailer in my country.... without basically *any* packaging except for a plastic bag... Run a whole disk write test - yep it was bad...
HDD's in my NAS instances lasts for about 35-40k hours. It can be ok to buy used one's, but there is one crucial thing- the transport of the hard drive. If the drive is transported incorrectly then it will be dead in 1-3k hours if not earlier.
9/10 if there going to die they’ll do it in the first few months of operation . I’m still skittish on 2nd hand drives. It’s not worth buying them now since new drives are so cheap.
RAID is not a backup. I buy used drives and although a few have failed, it hasn’t risked my data; One time I was setting up backup task to a remote server but instead of making the remote folder look like my local folder, I mistakenly chose to have the local folder match the folder in the cloud. The task ran and deleted my local files. ZFS Snapshots to the rescue and nothing was lost. The biggest risk to my data hasn’t been failing hardware. I have been the cause of most issues. [EDIT: English is hard]
Pretty sure I've bought a couple dozen drives on Ebay. WD Blue 1TB HDDs are my most common purchase. Never had a failure. Actually got 2 of them in the PC I'm typing this from. Building a NAS with used WD Red at some point, and I have precisely zero concerns relating to the drives themselves. SSDs scare me a little, but I've never had issue with those either. I've got a used NVMe in my laptop, and I ran my old desktop from a used Micron m.2, which is still reliably serving its new owner. Both purchased from Ebay.
I want to be able to do this. Which software are they using? I want to check my NAS storage,see how much life they have in them. NAS storage already does have SMART in use. But it only says its good or not. Not much in detail.
As someone who doesn’t know anything about what was said in the video. Why are people that use them for mining selling them instead of just continuing to use them if they are working still ?
9:00 Best bang for the buck right now seem to be at the 10 TB level, in used Amazon drives. 12 and 8 are very close, 14 pretty close, the BIGGER drives go up pretty fast.
My NAS has had 5 3TB WD Red for ages, all sit at ~75k power on hours and not a single UDMA_CRC_Error on one of them nothing else. Added a 6th to move to RAID6 2 years ago for safety but no issue in sight.
This isn't even remotely true. I mean, in some instances, maybe, but depending on your definition of "old" that is definitely not an always true statement. Two drives pushing several years of power on time and minutes away from failure is obviously about as good as printing your data onto sheets of paper and storing them outside. Remember, you will have to read all of your data, for it to be safe. It existing on a magnetic platter does you no good if you can't read it all. Rebuilding a RAID1 when one drive fails is the most stressful thing a hard drive can do, the chance of your second one failing during the task is astronomically high. Physical data recovery is much more expensive than reliable hard drives. You're talking thousands instead of hundreds. Don't be cheap unless your data isn't important. If you can't afford new drives, then this argument is moot, you're going to use whatever trash you can afford regardless of what is actually the correct choice.
@@flyingtentacle7631 New drives can have QC issues and fail within the first year of use. Older drives that are in good health and weren't abused are far safer to store files on in a mirrored setup than a single new drive.
@@randomman057 They are better even if you use one. There is like 10% chance that new from a factory hard drive just dies on you, within first year. Good used drive is already well past that failure stage, so it will be fine for a loooong time, before wear actually starts to show up.
@5:38 Chart is incorrect - Shows WD as having all 5 attributes monitored. But the video goes on to say and show that Seagate was the only one to have all 5.
I always buy hard drives from ebay, refurbished are amazing aswell, bought a ton of 4tb drives selling for $16, less if you buy more and they were fine with an offer of $12 per drive in bulk
I’ve done it just to put games on. Got a $20 drive and it kept my games just fine for a couple years before it started buggin. Pretty sure I’m the reason it broke too because I move my pc around a lot. Now I just have 3tb of SSD storage
I have a Seagate 2 TB HDD I bought used one eBay around 2012 that's still in my main PC today. It doesn't spin up too often these days as I have 2 TB of SSD storage, it mainly is rarely used mass storage like downloaded installers and mods.
I test a lot of hard drives (although I am not a data recovery specialist) and I recommend doing a full read and write scan on all new and old drives that you buy. Then review SMART. There are lots of utilities that can do this. You can also do it 'manually' in windows by doing a 'non-quick' format (full write scan) and a chkdsk X: /R scan (full read scan). This can be very time consuming but it is worth it. Best example "it was worth it" was spurious CRC errors on a batch of 6 brand new 4 TB drives (back when they were the bleeding edge capacity). If you return a drive with faults to a retailer within 7 days they will often bin it as DOA and refund or replace immediately.
Its nice to see some correct information on Chia farming (mining). With Chia farming, plots are written to the drive one time (maybe two ) and are very light weight. I would rather buy a chia farming HDD than a data center HDD or even a personal HDD as even personal HDD's are used more than Chia in a lot of cases
I am more concerned with the drive type than its use. Heavily used datacenter pulls have treated me far better than anything you can get from best buy.
@@andrewt.5567 Its just how hard drives work. If new one survived first two years - it will live for a decade or two. Spending these two years in datacenter is basically best reliability test you can do.
Sort of, you increase the likelihood of a drive failing, but you reduce the likelihood of your data being lost. (assuming you aren't using straight up striping with no redundancy)
I bought 10 3TB drives for about £150, 8 went into a nas as raid 10 (I wanted speed and redundancy) and I also have a new external 10TB drive used as a backup. This means I have 2 cold spares ready to go and saved alot of money on the initial cost with 0 drive failures in almost a year i consider it a win.
I have a several drives(HD103SJ) that are hitting between 70k and 85k hours. But they spin down when inactive? So not sure how much of that is actual active use on the hour count.
Even a new HDD is not worth risking your data. No substitute for backups. 3-2-1 and you’re probably good. I sleep at night with DVD-R and Amazon S3 deep archive REPLICAS.
Those prices for the 18 and 16 terabyte drives are always hella good 200 or $300 is amazing that's how much at one TB was when they first came out so that's pretty good amazing actually
This video should have pointed that shucking drives usually means you end up with a slower 5200RPM drive and a worse warranty (or a voided one altogether) than a regular internal drive. You save some money but you take your chances.
My old shipping co-worker flipped some drives on their side on a table once, like this video (3:00). I marked them as "will fail," they were dead about a week later. NOT knowing how their treated makes them almost worthless.
Shoutout to the writers and Alex on this one, really well presented! The chronology and dialogue was well written and easy to follow. shoutout editor too and film crew too.
I prefer new, but have gotten used drives without much problem as well. Refurbs are generally as good as new in my experience, or at least close enough
Generally they just don't last quite as long, but depending on use case could last for years. Stuff for servers or previously in servers is a bit different tho
At 5:39 there appears to be an error in the chart, if your words and the following images are to be believed. Chart shows Seagate and Toshiba as having question marks (Unknowns), but verbally you said WD and Toshiba. A few seconds later, it looked like you showed a Seagate having given all the attributes.
If you go the external drive shucking route and pick up a WD, you'll likely need to cover the 3rd tooth of the power connector on the drive for it to work internally due to requesting extra voltage (as a measure against such conversion). I had to do so with a 12TB ext I picked up.
This has nothing to with WD discouraging shucking. The SATA specifications changed in revision 3.3 (2016), they added support for Power Disable which allows you to remotely reset a drive. Usually only server drives have this feature but WD uses server drives for their external drives. Consumer PSU's haven't caught up spec wise to fix the issue so it continuously supplies power over that 3.3v pin causing the drive to be in a constant reset state. If anything it's a problem with the PSU's.
I've had the most luck with Toshiba drives, got a few with bad sectors, one of which has ~3500 of them, yet there are no slowdowns, no clicking, nothing, it's like there's no issue at all. Same with the other Toshiba drives, altho less bad sectors. On the other hand, WD and Seagate with bad sectors behave in many ways, from slowdowns, clicking, even grinding, some refusing to show up at all. So for me, getting a used Toshiba drive has never been an issue, even ones with bad sectors. The other 2, i try to avoid used unless i can test them first.
Another good thing to note is the quality level of the used HDD. You may find consumer grade options that are the same TB/$ as the used enterprise grade, but the difference in quality and longevity can be worth it. Though, that can also come with its own set of caveats as some enterprise drives run much faster and more power hungry than consumer drives that are more likely to be idle. Used enterprise drives can be great though, especially if you go through all the SMART data like you guys showed! I personally have a combination of new and used enterprise HDDs (I have more storage at my house than the large OEM I work for...), and have had great success with both. It's always great to see this sort of information clearly conveyed so that people that might be on the fence will have a bit more confidence with tinkering around with used hardware!
Smart data can be reset easily with program. I have for WD. Bad blocks? gone. Power on hours, zero. Good as new. Don't trust smart data, do full scan, even on brand new drive, can be surface issue. Smart does not detect the errors unless you try to use each block. Hdd does not have a background integrity check.
I’m of two minds here. On one hand, I guess you could store games on used hard drives to minimize the risk. But on the other hand, you’d be much better off storing games on an SSD for the sake of load times.
Meh, with the size of today's games, (COD looking at you..) plus they are only getting larger, a 250gb SSD new for the price of a 2tb used HDD to me isn't worth it even with a few seconds extra loading time. Plus as it's only the game data stored if the drive fails you can just buy another one and reinstall the game. Furthermore, if you're not running the system of the hard drive and it's purely the game data only stored you will find loading times are not as bad as you would think.
Be careful when SHUCKING a hard drive, some western digital will act dead because they are getting power to one of the power pin on the drive where the external case doesn't provide power to that pin, you have to bend that pin until it breaks and then plug it in normally
"... a dedicated NAS has a singular purpose." - I want to correct Alex on that. Synology NAS as show in the show as a comparison to the home made NAS is NOT only a NAS! It can also be: Webserver, email server, VPN server, WebDAV server, Database server, Home Assistant, iTunes server, DHCP server, DNS server, LDAP server, Git server, SSO Server, Chat Server, Home Assistant server (via Docker) and more. And THEN comes all the other standard file sharing features like, Antivirus, Clud Sync, Backup Vault, Media Server, Documents Server, Download Station, SMB and more.... The extra 722$ is counted for mainly all the hussle that one individual would have when he tries to install all these services and use them on its own home made NAS. Unless somebodys pure goal is to have a dedicated external data vault with file transfer capabilities via SMB or other methods, then heck yeah! Best sollution ever and fully supporting it. But as soon as somebody would love to have some nice to have features like getting rid of Googles monopolistic cloud services and move all the data to it's own, then something like Synology is priceless because everything is at place. Just click and voila. If you even have a dedicated IP address, you don't even need the NASs provided quickconnect or any DynDNS. So while I'm fully supporting the video and it is a great idea, the sentence "... a dedicated NAS has a singular purpose." triggers me a bit because there is the reason why it is 722$ cheaper to build you own NAS.
I bought used HDD most of the time... But I usually choose the youngest one, and choose the better one like the Red WD. I also asked the seller to test them with HDD sentinel / HD Tune too.
I just bought a lot of 9 4TB HGST SAS drives. They were like $15 per drive. They were manufactured in 2015, but had just under 40,000 hours on them (~4.5 years) I'm putting them through their paces right now with badblocks under Linux. So far so good, no issues. All but one had very low, or 0 levels of ECC correction counts in the SMART data. None had uncorrectable error counts. (Since they are SAS, the data output is different than SMART on SATA drives) I wouldn't use drives like these in a client, but in backed up redundant pools, why not? I just use more redundancy than I usually do. If one fails, I can get another $15 replacement. That beats spending thousands of dollars in drives when setting up a new pool.
I absolutely love HDDs and especially for my Hobby machines where no critical data is stored, these are fantastic 💪😎 In Germany there are always tons of used 2.5" HDDs with around 6€/250GB or 10€/500GB. All by companies with warranty. Can't go wrong.
I have only ever had 2 HD's fail. The first one was in 2005 and 2 years old. Lost the first 3 years of my sons family photos, lesson learned. The IDE replacement is still going strong, but only used for backing up backup and just 120GB. The only other was my first SSD, which lasted 3 years, but even out of warranty, OCZ replaced it with a upgrade. This PC has two Cex specials that i have had for 5 years and still going strong. You can buy used with confidence so long as you backup your important stuff.🙂
That reseller warranty and recertification process is most likely, Step 1: buy a laser engraver Step 2: claim you recertify drives and brand them as such Step 3: offer a warranty, so people feel safe. Tommy Boy said it best.
He said that there’s proof of recertification when he checked the drives. So their claims of recertification was correct, even for the drives that had listings that didn’t say they were recertified….were actually recertified. Buyers will notice and you won’t be selling them for long. It ain’t that easy anymore.
@@Gold63Beast not proof, indications that MAY align with recertifying. Could also just be double pass erasing or something like that.
@@sultanofsick yeah you May be right but like he said, reputable sellers are selling them and buyers are satisfied. So it doesn’t seem like sellers are actively trying to scam buyers. I think both buyer and seller know the deal, both know the trade-off that’s explained really well in this video.
But it says so ON THE BOX! 🤣
Ebay=Scammers, News? I think not.
An investigation into old, used SSDs, the prices, drive health, and pros and cons would be really useful.
Generally SSDs are much better, there's no moving parts, so the only thing to likely fail is a dried up capacitor, but that could happen to anything and the NAND itself can fail, but unless you're looking at QLC, most drives are rated at roughly 50-100TB written per 100GB, so aside from defects or farming crypto it would be pretty hard to kill the NAND itself.
The real problem with used ssds, particularly old ones, is going to be the lack of dram cache and they often aren't all that much cheaper than just buying a new one, at best you're looking at double the storage for the same price but on an older/slower drive for example I've seen the WD SN750 500GB go for $60 new and the SN550 1TB $60 used, that's not bad, but I've seen much worse like older sata drives lacking any dram going for the same price as those two newer WD drives.
If it is SSDs used to plot chia they will be thrashed. While these HHDs used to store the coins didn't do a lot of read/write the SSDs used to actually plot will have tons.
I personally wouldn't trust used SSDs for anything other than content you've got backups of or don't mind losing. As with any solid-state storage devices, they can just die without warning and you don't have the acoustic reporting to tell you if something is wrong.
If it's SATA, I've been getting the Samsung 850 Pro from 2016 or so (10 year warranty, so still valid!) and SM863, also from the same era. Those 2-bit MLC V-NAND chips from that time have an amazing endurance, worsening as they added more layers. The SM863 is enterprise so includes power loss protection capacitors.
As for reliability, all SSDs seem to have some troubling avenues for losing lots of data easily. I stick to just having the OS and applications on it. User files, media, etc. on an HDD. At least the enterprise SSDs probably have more robustness (underprovisioned more for more spare sectors, more solid ECC).
I see no point in buying used SSDs.. you can get new off brands dirt cheap; which do fairly well, and you don't know how much data was written either, unless specified. SSD do fail once they reach their limit. Plus they are hard to detect when they fail, can die any time; had some good brands die on me before.
Impressive how Linus had pass his skills of dropping things to everyone that host videos
And YT algorithm do their job well to show it to group of people who love falling expensive object
It is literally written in the script. They have a 'memeable moment' requirement, dropping things is probably one of them.
It's become a corporate policy. I'm going to assume the Toshibas were broken BEFORE the recording. (I have NAS drives with 75,000 hours on them).
*Used to work
job requirement. no fumble, no job.
Just an inconsistency note: the table at 5:38 indicates that the WD drives indicated all 5 SMART fields of concern, but Alex says right after that it was Seagate who covered all 5 SMART fields of concern, not WD. The screengrab shown when he says this seems to agree with Alex rather than the editor on this one.
Edit: lots to keep track of in this video, quite a good job overall
Good catch
I have an x16 and can confirm that it reports all 5 SMART data fields.
That serverpartdeals seller has always been pretty good in my experience. I've gotten a few 8tb wd enterprise drives from them, and they're always super responsive
How much??
@@ezra1369 I just got an offer accepted from serverpartsdeals for 3x 10TB drives @ $76ea - SAS drives tho, SATA might be differently priced.
@@theironangel767 SAS drives are meant for enterprise use so they're going to be built better than even prosumer SATA drives for NAS use. Frankly, this video really undersells the longterm reliability of used enterprise hardware. However, SAS requires a bit more knowledge in order to get working in any sort of setup.
Yeah, I bought 9 of the Exos X18 16GB drives recently. Had 2 failures they replaced, and replacements were good. Both bad ones had failed with uncorrectable sectors during initial testing.
@@randomman057 Sure does, 2nd hand enterprise is the way to go for a cheap build out. You can get some great deals if you buy irregular format drives and know how to reformat them back to 512b - as long as the drive supports changing their allocation, some are locked, especially SSDs. I got 10x6TB 520-byte drives a few months back for $45 per drive, low writes only a few thousand POH and from late 2018/2019
Chia is very lightweight on storage drives. So if the eBay listing specifically states it was dedicated to farm chia it is most likely better than the majority of other used drives out there.
Yeah it's actually better on the drives as well as the drives are kept spinning instead of starting/stopping over and over again
@@Schmicky96 Source is knowing how Chia works a drive compared to how a drive is used normally. It isn't a trust me thing, it is just how things work.
@@Schmicky96 The source is literally this video. They said the same thing.
@@Seris_ Holy shit, are you guys trying to be wrong on purpose? 😂
Crypto bros clearly don’t understand how an economy functions, so I highly doubt they know a single fucking thing about drive loading and wear.
Besides Chia is currently worth $28.78USD, so not only is your pretention hilarious, it also highlights how fucking utterly clueless you cryptobros are.
I've gotten some amazing deals on used hard drives. Got like 5-6 Firecuda drives that I'm pretty positive weren't even used. Payed like $10-$20 a piece for 1-2 TB drives.
Yeah I was about to get new drives for my nas but I just got a deal for 1tb drives at 11€ a disk shipped. Got 7. So it definitely can be great.
I have had bad experiences with used drives so not always great. 2 years ago bought 9 500gb drives, on arrival only 6 were working, got a partial refund (ebay is always great with that, even if seller is an ass, ebay will step in and find a solution) and now 2 of the 6 that were left are dying.
Bought another drive (to replace one of the two that was dying) and the thing literally came in a letter, it works and tests fine but I don't trust it now.
@@bradhaines3142 🤓
@@bradhaines3142 didn't even notice I put payed lol
@@imnota it can be hit or miss, that's for sure especially when you find them for sale really cheap. When they're certified refurbished and closer to the market price it's a safer bet but when you find a really good deal I'd basically always consider it a gamble.
@Jake
Out with the old in with the new. Everyone wants that high performance of SSD drives, brings new life to even old PCs, like going from HD DVD to Bluray 📀
If you do a follow up, I encourage you to cover usage of tools like badblocks and hard disk sentinel's reinitialize test for burn-in testing drives. In the dozens of new and used drives I have had, it has been very reliable at catching early failures before the ebay money back guarantee/newegg return window closes. None of them that have passed this test have failed before they were re-sold years later, personally. Cheers!
HDS is an amazing utility although it does take a long time to learn all the nuances of its different test methods, what types of errors they expose and what they don't. Essential if you care about your data storage or spend time fixing computers a lot.
Wow! thanks for the tip!
This 100% badblocks is great, that and something I wish they would of mentioned would be about S.M.A.R.T info tampering
@@mubok3743 I wonder how many sellers are experienced to wipe out the SMART data, and reset them back to factory-new-like condition. Because its quite easy to fake a new hard disk if you reset its SMART counters.
Used to work at a little computer repair shop. We had all kinds of hdds from SSD upgrades that people didn't want. One time a youngster came in with $3. We have him a box and said he could take 10tb he was so happy
Ten terabytes for $3, good deal!
and here i am looking at 8-10tb drives for £120-150...
Great
I bought a 4TB WD Gold HDD on eBay about 4 years ago for £60. It's still working perfectly. I risked it because it was reasonably cheap and as it's a WD Gold, it's intended for datacentre use so it's designed to last long (MTBF of 1.2 million hours!) and is less susceptible to mishandling. Buying a used hard drive isn't always a bad thing.
Edit: I just checked how long my drive has been running for and it's currently at 54,480 hours, SMART status is all good.
For Chia, just writing the final plots to the drives doesn't wear them too greatly, but if they were used as the temp drive to make the plots, then the write amount goes up exponentially.
Lol no one uses hard drives to plot chia... it's literally not even economically feasible to do so
@@quantum8522 it's not as bad as you think it is. I didn't have the budget for high capacity SSDs coz they're expensive here, but I used multiple HDDs each with its own queue up to what my RAM allowed and they did okey. It wasn't economically feasible when people where farming solo and trying to beat the growing network space, not to mention the declining chia price.
@@quantum8522 it is possible in raid SAS drive, but Ramdisk is even better and not harmful as SSD
I lucked out on some WD Reds that had chia plots but the price was 1/3rd of the original and they had like 300 days total work. For me it just passed the infant mortality stage and it was already pretested lol
Every "factory refurbished" Seagate drive I have bought from ebay had its SMART stats, including power-on hours, reset. Buying a used anything is always buyer beware. But I have actually received several brand new DOA drives while all of my ebay finds are still working after years of 24/7 operation, so your mileage will obviously vary. RAID6 or z2 is a must if you care at all about the data regardless of whether the drives are new or used....
I have a 30gb HDD from way back in 2008, that still works relatively well. It's used right now as a storage for some games like XCOM and other, older, titles. It's being read and written to somewhat infrequently, as it's not a boot drive for anything, but still, kind of amazing that it still works.
But why. A 30gb hhd isn't worth the space it takes up in the case
@@leprechaunbutreallyjustamidget If it still work why should he/she get rid of it?
@@leprechaunbutreallyjustamidget If he has space it available why not use it? He already owns the thing.
@@jamesbutson6347 power usage
@@leprechaunbutreallyjustamidget That HDD was the first one i got as a present from my dad. Still kinda keep it as a memory, pun intended, along with other things i got from him. I do have a normal 1tb M.2 drive on the motherboard (got it on the cheap, as the company that sells PC parts went out of the business, and were now offloading the leftover parts) and a 2TB additional storage HDD, so i'm not putting valuable files on that ancient hard drive. Just keep it around for the memory's sake, as that little present opened up a whole world of PC building for me.
I grabbed a whole bunch of Seagate x16 drives that were ex-chia mining for all around $200 a pop, bargain.
Very glad you guys did this video, helps educate what to look for.
Fortunately my seller did amazing packaging, but another issue here in Aus is drive on Amazon shipping from US basically loose in a box. Really bad.
Oof. I have how Amazon doesn't ship their items with care. Hope you get your drives in good condition.
Even on the fast internet I have, it can take over an hour to download larger games. An inexpensive HDD is a pretty decent solution to keep games stored locally if you are not currently playing them and want to keep the ones you are playing on your SSD.
A Steam Cache is useful too. It's like a separate little server that you download games locally from instead. Pretty neat, especially if you play a lot of AAA games.
only one hour to download larger games? That's...insanely fast already. Took me 11 hours to download Cyberpunk for example lmao.
But yes, having a large HDD to always have all your games ready is a must. Using a used one too
@@prnzssLuna yeah, it took me 7 hrs to download BF1. I wish I could download that fast.
@@prnzssLuna took me a couple of days lol
@@reptilez13 Steam Cache is still slow as shit tbh. I run a pc cafe and we used to have a steam cache running, but even at 1gb/s (the limit you will hit on most affordable multi-port network switches), very large games would take 30 minutes or longer to transfer over and most people like to try a handful of games before settling on something. We've since migrated to using 1TB SSD along side a 14TB HDD internally for each station. The SSD and HDD are tiered, so the most played games basically stay on the ssd and even if you choose something less popular, at most you'll have to wait through a HDD loading screen once, before the game moves over to the SSD automatically while you play.
Waiting a few seconds more for a loading screen is much better than waiting half an hour for an entire game to transfer over. Even manually moving games from HDD to SSD is faster than the 1gb/s bottleneck you hit with a network. People consider hard drives to be super slow, but high capacity hard drives can easily hit 200MB/s or higher when they aren't running an OS and the workload is mainly sequential.
Having an SSD cache and 10Gb network infrastructure would have also been a solution, but even that wouldn't have been as fast as our current setup and would have cost 10x as much. We would be better off putting 10TB SSDs in each station and calling it a day.
Adding to the anecdotes: From my work experience enterprise drives rated for 5 years usually hold up at least 50-60k hours and I'm currently emptying some drives on a personal server that have 72k hours and still run just fine(the server just got superseded). BUT some hard drives just die in the span of a year or two. It really just depends on luck more than anything.
I got a Toshiba 16TB Enterprise helium-filled drive through Ebay and it had just a couple hundred hours shy of one year's worth of hours and only something like 20 on/off cycles. It was perfect! (still is). Quiet and super-fast.
I liked it so much I ordered another one and never got it to be recognized by the computer. I wrote them and they wrote back and asked me to send it back... super easy. I got reimbursed and ordered another form the same people and it went in just fine and works as well as the first, but this one had about a year's worth of hours and 24 power-cycles. It is as quiet as the first one. All three were sent pretty well packaged with sealed ESB bags and foam around them.
I also got some SSD's (4TB Samsungs) for about $60 off regular price, but 'open box and 'tested'. They came in envelopes, but packed with cardboard, ESD bag sealed, no label or original box as they must be kept for records. One power cycle and 67GB written and erased, otherwise, just like new ones. And since these things might last for decades, I figure as long as they are in good shape when I get them, they'll probably stay that way.
The key thing here IMO is the power cycles. The fewer the better, and most of these resold enterprise drives haven't even gone through 1/10th of their MTBF rate. Chances are the drive will outlast your usecase and you should be RAIDing them anyways.
I six drives from GoHardDrive like 5 years ago and loaded them up with ISOs in an array. I've since sold those drives and they are still chugging away in my friend's server without issue.
Same case here got couple of WD Reds for cheap filled with plot files. Power on time 300 days and 30 power cycles still work till this day. I have read that chia storage drives are actually the best second hand drives you can get because they don't see much work after the initial writing
An important warning you missed, particularly in relation to shucking drives - if you plan on using any ZFS based OS (Xigmanas, freenas etc) DO NOT use any SMR drives, which are commonly used in those portable hard drives that people shuck, which is why they're cheaper in the first place.
Can you elaborate why? Or at least point to what I should google to understand it? 🤔
I wouldn't use SMR for much of anything other than long term backup.
@@snowhusk SMR stands for "Shingled Magnetic Recording". As the name implies tracks on SMR overlap like shingles. Initial writing on SMR drives is pretty reasonable, but as soon as you delete files things start to get unbearably slow since it has to move any tracks that are overlapping on the file you're moving/deleting first. I've started buying them for long term backup/storage in place of WD green drives because the price:byte ratio is exceptionally good, but again I wouldn't use them for much else than files you don't intend on *ever* moving around or deleting.
@@s01itarygaming thank you
I can’t wait to see how many he drops them this time
@Astral lmao
anecdote time: I bought two used hitachi 2tb drives with ~10,000 hours on them like 10 years ago and still am using them. Now they have ~92,000 hours. Idk if i was being stupid, but I actually bought them with the mentality that if they were still working after being used for a while, then they probably weren't about to fail.
I got a refurbished 4tb one for $60 and according to crystaldisk, it wasn't even used. Has been working great to keep big games backed up.
Did the same with a decertified 8tb easystore. Used crystal disk and it said it had like less than an hour of use and a handful of power ons. All for less than $100! Great and cheap, perfectly fine for mass storage!!
Oh those drives usually have their SMART data wiped. 1 hours powered on is a red flag. That said, there's nothing wrong with throwing them in a raidz2 array. Even with redundancy they're much cheaper than new drives
@@NavinF oh wow I had no clue haha. Good to know!! Thankfully I’m just using it to store footage and photos from projects/events I’ve finished editing, so I’m hoping it holds up over time! Thanks for the heads up!!
They basically reset the Smart sensor.
@@syarifairlangga4608 There is no "SMART sensor." The drive keeps its SMART data on a normally inaccessible area on the platters and it's not impossible to clear it
IMHO, if you have data to archive (not back up), let me repeat, archive, then WORM storage is the ONLY solution. Enterprise often uses LTO, but I have found that DL BD-R is great for the average consumer. I have several TBs of data archived to DL BD-R, and I fully anticipate that each disc will remain readable for at least 20 years. Of course, burning, verifying, cataloging, storing, and reading is tedious and time consuming, but too many people treat data as an afterthought and that sort of carelessness is why data get corrupted and lost. People only care about their data once it goes poof. Learn to care about your data before it goes poof so you'll invest the time and money to ensure it is secure. Once you've done that, failing hdds will be the least of your worries.
I got curious when Alex started talking about power on hours, and the expected lifetime of drives being 45,000 hours, and setting a ceiling of 25,000 hours for purchasing used drive, so I opening up crystal disk info and took a look at the power on hours of my drives. I was absolutely floored to see oldest drive in my system, the one that I use for the majority of non critical software installs, has a whopping 70,000 hours.
I bought 8 x 3TB HGST Sata drives via Ebay in 2016 and set up my home server. Those all had about 20K hours each and were running 24x7 from 2017 until spring 2022 when I changed my server setup, no issues, no failures. New setup is 4x 10TB HGST SAS drives, ebay for $80 each, 12-14k hours each, running flawlessly since I set it up about 6 months ago. For the price, I keep, depending on the box, hot spares or a cold spare handy just in case. Far more effective from a cost standpoint even if you happen to have a drive die.
Not even used drives, I've sometimes found brand new ones dirt cheap. I've seen WD enterprise grade drives for less than £10. Sure they're only a couple of hundred gigs in some cases. It's also worth noting that you can't necessarily trust shucking some external hard drives, especially the ones meant to be accessed remotely, as they can be locked to only work with the controller boards attached to the cases. Another thing worth noting is that you should only ever use the numbers provided by programs such as CrystalDiskInfo as a rough guide. I've seen conflicting reports between different programs, some drives don't even report some numbers, illogical Power On Count vs Power On Hours, POC and POH numbers even for brand new sealed drives, and it's possible to tamper with the firmware to report less than is the actual case.
Like Alex said in the video: Ebay protection is very good for consumers. Make sure you read, and fully understand the policy before you buy anything. that way if you end up in a situation where what you receive doesn't match what you were told you were supposed to receive, you can tell the seller "your listing says no return, but ebay's policies say this, so i'll be returning this, send a shipping label" and their options are do it, or you go through ebay and get your money back regardless.
Sellers indicate "no returns" as a deterrent to flaky buyers
Great video. I recently bought 4 4TB WD Reds for my NAS from eBay (taking it to the max size, stinkin' 4TB limit), supposedly former pulls from a NAS, and while I didn't check beyond a SMART Short and Conveyance test, the attributes were fine, with no ATA errors in the log. Paid $30 a drive when they're easily quadruple that cost for a new one.
As long as you're running in a RAID mode that can survive 1 or 2 drive failures, you'll be fine.
For full on failures like yours Alex, look into RT Tools's R-Studio, amazing recovery tool.
This exactly. As long as you aren't buying decade old drives and purchase them from reputable sellers you probably won't have an issue.
Used HDDs are perfect RAID 5 or 6 configurations. It's definitely cheaper when the cost of replacing a single drive is less than half the cost of new.
Buying used in bulk leads to crazy deals. I've regularly seen 10-20 or more 1TB drives selling for under $10 per drive, sometimes as low as $4.
Hard to go wrong with savings like that.
So, I purchased over 60 used 4TB hard drives from eBay for Chia farming (no it hasn't been profitable). It's been about a year of continuous use. Very few have failed! It's hard to tell which have truly failed and which has lost connection due to USB dock issues. Install is under a bed so it is not easy to dissect. I lost connection on a number of them in the early days but for about 9 months now, farm size has been consistent.
Majority of them have 40-60K hours on them when I got them. Some are WDs all 7200, some are Hitachi 7200, some 5400, some are Seagate 7200, some 5400. For unimportant storage, it is definitely worth it the reduced cost/TB and reduced carbon footprint of not having new drives manufactured. Though, it does make you wonder about the people unable to buy the drives I bought, if they ended up buying new negating this benefit?
A note, I also have a huge Distributed Computing farm (Linus Tech Tips team) and 1 of the drives in my PCs (labelled Haswell#11 in BOINC) is a WD2502ABYS Enterprise grade drive with 86.5K hours on it. SMART report is perfect! Some drives are engineered really well!
I say that power cycle count is very important to judge drive wear. Temperature cycles causes wear, whereas if it's only had like 25 power cycles, you know it's been at a consistent temperature through most of its life. Almost all my drives do have a low power cycle count.
This video made me check, I still have a usable hard drive with 45000+ h of power-on time. its a slow as shit 5400 rpm that i have kept migrating between systems for years and that has seen very little to none usage in the last couple of years (hence the fact that it still works at all), but still, its running in my system right now and I recently used it as a buffer to move data between ssds without issue.
What do you mean by using it as a buffer?
@@dadodi I'm assuming it's making a quick back up of stuff when moving files to another drive. So so you're transferring from SSD 1 to SSD 2. The aforementioned hard drive would serve as the back up buffer so there are multiple points of failure. It's a good and simple way to do things. However, with this method, it will take a little more time.
I have a few sitting in the 40,000's hours. The 'champion' however is a 5400 RPM drive in a work PC that I built which is nearing 10 years of spin up time. All its SMART parameters are OK but I'm hesitant to do any testing on it all. One day it will be layed to rest but not today. I think the very lightweight platters in these slow 2.5" drives gives them an edge into living a long life!
@@indoctrahol that too, but i was mainly using it as a way to temporarily store the data because having the 2 ssds in the system at the same time wasn't an option because of limited m.2 slots (and no other drive having enough free space)
45000h isnt a lot for hard drive actually. About half-life if you didnt get faulty one, it will survive AT LEAST just as much.
Note on shucking, do not buy the Western Digital Elements line for that. The logic board on those drives is USB, there's no SATA port on it; if you want to use that inside a computer with a USB 3.0 header to type A or C or whatever port and hook it up internally over USB it'll work (in fact if you want to do that because you ran out of SATA ports, maybe _do_ do that), but not in a slot loading bay that hooks directly up into a SATA connector that's permanently mounted like in a standalone NAS enclosure.
I did not learn this from experience in frustration trying to do that, but I was considering shucking one myself thinking it was a replaceable drive inside an enclosure like the My Passport line, and learned about its contents while looking for instructions on how to open it without destroying any and all plastic clips holding it together.
Not true only some are but you should be researching anyways before you buy because same model could have different model hdds some faster than others etc some with helium etc
@@primus711 I probably should have specified the 2.5" drives. The current line are entirely USB drives.
I bought a "new" hdd on ebay awhile ago. I checked the crystal disk info for it and it had been used for like 2 years. You really got to be careful with buying hdds.
2 years is probably fine tbh. It's not like drives these days aren't built like tanks mostly.
@@gorkskoal9315 Unless you mean toshiba ones. It seems they have a tendency to early failure, but just reliable as anything else later (and better then seagate, so there is that).
Hi Alex, I just checked 2x WD disks though via USB as it's external. Linux Mint -> GSmartControl (have to install but it's in default repo) -> Attributes Tab shows no LBA, but click on the next "Statistics" Tab and there it is: "Logical Sectors Written" and "Logical Sectors Read", both with Values.
So it seems WD disks are collecting LBA Written/Read but shows them somewhere else.
I have bought numerous used hard drives through the years. Never one problem. I've only bought from sellers with good rating. I'm from Norway btw.
If you buy an used hdd, let's say from Sweden for example, how long does it take until it catches a torsk smell on atomic level? Does it require to be wrapped inside an ylletröja to make it operational?
I'm from Finland btw.
I wish I had some Norway cold. I live hot smelly swamp Florida.
Me from norway
@@Keepskatin There are plenty of Norwegians that hate wind and colder temps, but I actually enjoy it very much. There's something refreshing about it.
100% same here. All my hobby machines run on cheap refurbished 6€/250GB 2.5" HDDs and NO issues EVER! 💪 Germany here 🇩🇪
I got a "refurbished" 12TB drive from Amazon the last week for $125. But according to the SMART data it was brand new and that was the first power up. Not sure what's up with that. The drive came in a static-proof bag inside a barely padded envelope. When the delivery driver dropped it on the front porch, I heard it hit the cement from my home office next to the front door. Been running without issues for about a week now. I basically copied all my old drives onto this one drive and I still have the old drives, in case anything goes sour.
@@petedavis7970 Beware that for some models the Smart can be reset, even if it had bad sectors. I did this to WD drives. There are utilities but hard to find and are paid. On old hdd, I would advise to write some data or just read with Hdd Scan and keep it offline for a few months and read again and compare the timings for sectors. If it gets high timings, like 70ms, 120ms on many sectors, it means surface is degrading and got weak and if you write data there and don't overwrite it in a year, it may be unable to read it back (weak sectors).
Note that even new hdds can have this issue, weak surface, so new is not always better.
There is a utility DiskRefresh, which reads and writes back each sector, to remagnetize the area. You can use that from time to time, like 6months, to prevent silent corruption (bit rot).
As long as one is able to find sellers who regularly deal in white label hard drives, with their own warranty and customer service to back it up; they can be a relatively safe option for maximizing storage capacity for those who wish to keep their data local.
Amazing how often the warranty coincides in length with eBays money back policy. Literally no real downside to offering returns for faulty drives for 30 days when eBay basically does for you, but without the return if they say so.
I mainly use decommissioned hard drives for cold storage backups or for non-critical use, like a Plex server. When I get it, I run a thorough test of the drive, then copy a bunch of data using a USB 3.0 drive dock. When full, I put them in a storage case and store them on the shelf. Can't accumulate power-on hours if they are offline, nor can they be encrypted by hackers. I also use LTO tapes, USB thumb drives and SSD drives so I have multiple copies of important data on different media for cold storage backups. I recently picked up a 20TB rebranded Seagate drive for $199. Tested fine, copied a little over 19TB's of data to it and now it's in a storage case on my shelf. Multiple backups on different media, stored onsite and offsite is key.
Could you guys do more videos to explain certain aspects of tech? Like the different kinds of ssds and what use cases they are best for? Or the different kinds of panels in Tvs and monitors, i.e. tn vs ips vs oled vs quantum vs whatever else?
I know I can google all this eventually but I would never remember it all anyway and it would be sick to have a comprehensive video that I could reference about like each category of tech. You guys do pretty good work covering what you cover I just wish you would stop making sub 15 minute videos. Idk if you tend to do this because it is faster for production or the TH-cam algorithm tells you this is the best way to make money but I follow this channel to learn. So don't rush the video lengths, you don't need to, I'll eventually watch it all!
We have a whole channel for this called TechQuickie
@@LinusTechTips 🤌
I've built a bunch of sketchy servers with raid 6/5 arrays using old drives and *knocks on wood* they've yet to let me down after over a year of 24/7 service. IF you use a raid array of older drives, one helpful bit of advice is to always have 1-2 extra drives set aside for each machine to quickly replace one of the active ones if it happens to die.
Raid is not a backup and as long as you have a contingency plan, I see having a bunch of older drives that I feel ok discarding at the first sign of failure to be a very safe bet.
Great information as always. The only thing I would add is that Amazon sells the WD enterprise drives refurbished and they do accept returns within a reasonable amount of time. I have had no issues with the “Renewed” drives as they label them.
SAME!
Full time chia farmer here. Chia doesnt really hurt the drive. Once its filled, it might get a read once every few minutes. Theres higher risk storing the hdd on a desk since you camn drop/ bump it/ break it easily.
I just want to say, shucking often involves having to deal with SATA pins that need to be cut or taped off so that they'll work as internal drives.
I know it's not always available these days, but I find it easier just to use the molex connector with a SATA adapter
The best way to get _stupid_ amounts of cheap storage on eBay is to get old SAS drives. They are often dumped to eBay in bulk _way_ before they are actually used up, they're usually 10k to 15k ROM drives, they're usually _ultra_ reliable, and they are sold for so cheap because they are usually formatted in 520 bit sector sizes and require a SAS controller to use. So, what this means is that you will need a special controller card for them and then you will need to do a block reformat to get them back to a 512 bit goodness that most home PC's can deal with. The controller cards can be had for pretty cheap on eBay, but you do want to make sure that you get a model that can have the firmware flashed to be better for non-enterprise use. And block reformatting is super easy once you figure it out and you can queue up as many as you can plug in before going to bed or work or whatever.
And from what I've read, what happens a lot in enterprise data centers is that drives are basically _never_ run until they actually start hitting their r/w limits and instead often just get thrown out and replaced whenever they get full or close to full, just because hard drive failures are so catastrophically bad in enterprise settings that it's just not worth even the miniscule risk of trying to squeeze extra use out of them plus it's just a pain in the ass to try and keep track of which drives are oldest and need to be replaced, so when they increase their storage they just replace their existing drives with higher-capacity drives rather than add more drives on top of existing hardware. So they dump these drives that are already rated for super-high reliability when they still have about 95% of their useful life left. Definitely worth looking into if you're like me and you have a lot of storage needs and/ or want to have a private cloud storage server and you're willing to tinker.
P.S. it's also worth looking at old SAS SSD's to use as high speed cache.
6:10 Bad backplanes can and do cause corruption. I've experienced it myself more than once, stuffing known good drives into cheap 30-bay NAS cases and getting hundreds of errors in literally minutes. Data was irrecoverably lost, because ZFS and similar file systems will try to correct the errors by overwriting them with data from the mirror, which if it doesn't itself come back corrupted, may write things out corrupted.
IIRC Linus himself experienced this exact issue a few years ago with his "storinator", it's a bit weird to just dismiss the problem like that.
11:14
Warning on external drives meant to be used for backup.
A LOT of them (likely a majority by now) WILL BE SHINGLED DRIVES.
I've had 4 external 8 TB drives - 2 were Seagate and turned out to be ARCHIVE models, 2 were WD and turned out to not be SMR (but they WERE advertised as "GAMING" externals, not something like the MyBook backup drives).
Yeah, poor things. Get thrown and shaken around all the time, instead of cozy case or rack.
@@alexturnbackthearmy1907 Why do you ASSUME that singled drives didn't get mounted in a case or rack?
Both of my Archives have been in racks for YEARS now (it was cheaper when I got them to buy the external, then "shuck" it and toss the parts I didn't need to use).
@@bricefleckenstein9666 Huh? Are you sure we are on the same page?
@@alexturnbackthearmy1907 The only way for drives to get "thrown and shaken around" would be if they're NOT in some sort of a case or are in portable cases.
@@bricefleckenstein9666 Exactly. I think you misunderstood.
I have bought a cheap external hard drive from suspicious sites that when shucked contained a failing second hand drive
@4:30 I'm pretty sure this is how Toshiba would normally ship items out considering their "quality control" and life span... (Same with HGST). Terrible spinning drives IMO. That goes for both Enterprise and Home Use.
I wanna build or buy my own NAS. But the single most probably main reason I'm hesitant to do so and still using cloud storage is the AI capabilities of Google photos. The main reason I opted for cloud storage is that whenever I take a picture it's set to automatically be backed up to Google photos and I know I can find alternatives to initiate automatic backups. But the search feature in Google photos is irreplaceable to me. I can search and find almost anything in my photo collection. I can search for specific people, places etc.
Is there a way to search and sort everything in my NAS like I do in Google Photos.
Synology Photos
@@davyweng Does it work like Google Photos?. The search and everything.
Should be noted that when using external drives they will be slower and have a firmware optimized "safer" operation, as there usually "B" or "C" grade drives from the factory as not all drives come out the same from the assembly line. I have to do data recovery on one or two a year because clients use them as there main work drive, they usually just last about over a year when actually used every day
Gee! [in a tone of puzzlement] Clients using external drives as their main work drives is a dicey thing to do, given that they could be misplaced, dropped, or stolen in the workplace.
All that _read/write_ on an external drive? They weren't built for that.
Those external WD drives are great for the price.
I made an 8TB backup NAS using one and a Raspberry Pi 4. Sadly, getting your hands on a Pi is near impossible these days but you could easily use a Radxa Rock 4, whilst a bit more expensive(£60 instead of £40) it'll do the job just as well.
" Sadly, getting your hands on a Pi is near impossible these days " - Ask Jeff geerling if you can have one of the 20,000 he has.
@@Nalianna It's fine, I need 2 and have 2. I'm just saying that getting one now would be difficult.
@LTT - just a fyi, all the "Storage Based" Mining coins are worthless on HDD's. Potentially you could use ZFS and keep you ARC on SSD's to overcome the issues with this and flush the data to HDD's. But, you would still need relatively few HDD's, and a excessive amount of SSD's to make the bulk of it. As it requires fast, no-latency storage to function. (In that way, its significantly worse then GPU mining)
So while its remotely possible some one made a huge array of these drives to try and overcome this issue with a massive CPU and a Abundance of HDD's.
The more likely scenario is, small to medium businesses (and former small-scale datacenters) are dumping their old drives and matched spares onto the consumer market, after either replacing them or the company preparing to liquidate.
-- Here in Australia, we see this a lot. Ever 2-4 years, there is a massive influx of HDD's on Ebay and Gumtree. Most of these are smaller businesses refreshing there storage with newer sku's. As they cannot get replacement spares from the same series.
In IT as im sure your aware, Matching your drives is super important to provide predictable performance and reliability. Thus having hot and cold spares from the same SKU lineup is a absolute must. | Given the lifetime of such drives being around 3-5 year max, its often after 2 years you see 2yr power-on drives dumped and at 4 years a mixed 2yr power-on and their spares dumped on the market.
As to why the consumer market, and not liquidation outlets/auctions. Thats the simplest question. Profit. Auctioning HDD's in Lots, results in the overall unit price being low. Also, without consumer protections, less individuals (Datahoarders/HomeServer) and businesses are interested in investing. So you see these go for minimal amounts. | So unit prices on Ebay means top-dollar for your used (and unused spare's).
Perfect example of this is in 2019 i acquired 32, 8TB Drives for $17/aud and 24, 12TB drives for $8/aud from two separate Grays Auctions. | from this 30% of the drives had never been powered on. | my savings? around $6000/aud, compared to retail.
You should do a video on used enterprise SSDs. I think they might be worth it because the good ones use MLC flash which is superior to TLC and QLC found on most consumer drives. They also aren't as suceptible to physical damage during shipping.
For enterprise drives the ones with capacities like 400 and 800 GB instead of 480/512 and 960/1TB are better as they have much better wear leveling. Even better if you can get SLC ones.
So should you check your own drive's operational hours to see when to replace them? And how would one go about doing that?
If you have tons of money, sure... If not, I recommend some sort of RAID redundancy setup and monitor them for read and writes... ZFS is very critical and will let you know when they start to notice errors, at that point you swap it.
The real question is, how paranoid are you, and how likely are you to lose data, I went with 2 NAS boxes to satisfy my paranoia after my 1st NAS box which was a RAID 6 at the time, had a harddrive failure and a software issue that kicked out 2 drives, leaving me with no redundancy during rebuild.
This is a good video idea. I'd like to see what they recommend for a good way to pass over data safely.
10:44 That shunted external hard drive, is a Western Digital Elements.
The HDD in those do not have any cache whatsoever.
Bought a wd green ssd from a car boot sale for £5, turned out to be dead.
But I checked the warranty status and it was still under warranty so I sent it in for RMA and got a new one in return.
OG move 💪😎
2:14
Storj does not "mine" on a hard drive.
It is a BACKUP setup, similar to a low-budget BACKBLAZE, and those drives often see NO activity for months once they fill up.
Tends to cause even LESS wear than something like Chia.
I think testing AliExpress HDD's or even better SSD's would be interesting.
I've seen many high storage SSD's for dirt cheap (2TB M.2 for 52€) and im wandering if they're any good.
I've seen 16 TB or even 128 TB(!) SSDs from China on sale for $60. They're absolutely scams and do not work.
they are probably as good as the 10$ 1TB micro sd cards on ebay (always including the small cheap ass looking micro sd to usb adapter)
Sounds to be good to be true. I picked up a 1tb name brand re certified nvme drive on black Friday for less than 70 usd.
These TB are fake most likely. Write a bunch of labeled files on one and watch them disappear once you reached true drive capacity.
Imo: New SSD, keep personal stuff on that + backup on external, cheap high capacity ebay drives for games storage and just backup your screenshots if you care about those, can't go wrong! At least it isn't a quantum bigfoot drive..
I've suffered the drives sent wrapped in the thinnest of bubble wrap from Ebay. They are the worst, but as you say at least your protected. Selling any kind of IT kit on Ebay, I always overpack to protect the item and take a picture before sending. Apart from the odd nutcase buyer, never hadxan issue. Really enjoyed the video.
Got shipped a RMA return discount drive from probably the most trustworthy retailer in my country.... without basically *any* packaging except for a plastic bag... Run a whole disk write test - yep it was bad...
HDD's in my NAS instances lasts for about 35-40k hours. It can be ok to buy used one's, but there is one crucial thing- the transport of the hard drive. If the drive is transported incorrectly then it will be dead in 1-3k hours if not earlier.
I’ve bought well over 25-30+ used hard drives off of eBay, and I’ve never had a single issue.
100% same here. All my hobby machines run on cheap refurbished 6€/250GB 2.5" HDDs and NO issues EVER! 💪
9/10 if there going to die they’ll do it in the first few months of operation . I’m still skittish on 2nd hand drives. It’s not worth buying them now since new drives are so cheap.
RAID is not a backup.
I buy used drives and although a few have failed, it hasn’t risked my data;
One time I was setting up backup task to a remote server but instead of making the remote folder look like my local folder, I mistakenly chose to have the local folder match the folder in the cloud. The task ran and deleted my local files. ZFS Snapshots to the rescue and nothing was lost. The biggest risk to my data hasn’t been failing hardware. I have been the cause of most issues.
[EDIT: English is hard]
Pretty sure I've bought a couple dozen drives on Ebay. WD Blue 1TB HDDs are my most common purchase.
Never had a failure. Actually got 2 of them in the PC I'm typing this from.
Building a NAS with used WD Red at some point, and I have precisely zero concerns relating to the drives themselves.
SSDs scare me a little, but I've never had issue with those either. I've got a used NVMe in my laptop, and I ran my old desktop from a used Micron m.2, which is still reliably serving its new owner. Both purchased from Ebay.
I've bought 300ish in the past year.
All stayed functioning.
Just made sure to check out some details in the listing beforehand.
Not bad.
I want to be able to do this. Which software are they using? I want to check my NAS storage,see how much life they have in them.
NAS storage already does have SMART in use. But it only says its good or not. Not much in detail.
I mean I'd usually just go for the sketchy eBay white label drives
As someone who doesn’t know anything about what was said in the video. Why are people that use them for mining selling them instead of just continuing to use them if they are working still ?
For a mechanical hard drive the amount of power cycles is more important than power on hours, if those hours were not abused.
9:00
Best bang for the buck right now seem to be at the 10 TB level, in used Amazon drives.
12 and 8 are very close, 14 pretty close, the BIGGER drives go up pretty fast.
4:12 don't think the editer got the memo
My NAS has had 5 3TB WD Red for ages, all sit at ~75k power on hours and not a single UDMA_CRC_Error on one of them nothing else.
Added a 6th to move to RAID6 2 years ago for safety but no issue in sight.
Remember: using 2 old drives in RAID1 is less risky than a single new drive.
This isn't even remotely true. I mean, in some instances, maybe, but depending on your definition of "old" that is definitely not an always true statement. Two drives pushing several years of power on time and minutes away from failure is obviously about as good as printing your data onto sheets of paper and storing them outside.
Remember, you will have to read all of your data, for it to be safe. It existing on a magnetic platter does you no good if you can't read it all. Rebuilding a RAID1 when one drive fails is the most stressful thing a hard drive can do, the chance of your second one failing during the task is astronomically high. Physical data recovery is much more expensive than reliable hard drives. You're talking thousands instead of hundreds.
Don't be cheap unless your data isn't important. If you can't afford new drives, then this argument is moot, you're going to use whatever trash you can afford regardless of what is actually the correct choice.
@@flyingtentacle7631 New drives can have QC issues and fail within the first year of use. Older drives that are in good health and weren't abused are far safer to store files on in a mirrored setup than a single new drive.
@@randomman057 They are better even if you use one. There is like 10% chance that new from a factory hard drive just dies on you, within first year. Good used drive is already well past that failure stage, so it will be fine for a loooong time, before wear actually starts to show up.
@5:38 Chart is incorrect - Shows WD as having all 5 attributes monitored. But the video goes on to say and show that Seagate was the only one to have all 5.
I always buy hard drives from ebay, refurbished are amazing aswell, bought a ton of 4tb drives selling for $16, less if you buy more and they were fine with an offer of $12 per drive in bulk
I’ve done it just to put games on. Got a $20 drive and it kept my games just fine for a couple years before it started buggin. Pretty sure I’m the reason it broke too because I move my pc around a lot. Now I just have 3tb of SSD storage
@Tao
Once again, new person who buy ridiculous amount of HD capacity. What's your purpose.
@@johnsherby9130 Use eternal drives,.more reliable than internal and doesn't constant get used, working parts wear down in constant use.
@@Keepskatin tons of lossless videos and a huge steam library, plan to build a petabyte nas soon so this is minor, wanted a caching server aswell
@@tao_ikry9819 That's a lot of pirating.
I have a Seagate 2 TB HDD I bought used one eBay around 2012 that's still in my main PC today. It doesn't spin up too often these days as I have 2 TB of SSD storage, it mainly is rarely used mass storage like downloaded installers and mods.
I've purchased 240TB worth of used and refurb drives over the past 2 years and the only problem I had was that 1 was DOA which the seller replaced
I test a lot of hard drives (although I am not a data recovery specialist) and I recommend doing a full read and write scan on all new and old drives that you buy. Then review SMART. There are lots of utilities that can do this. You can also do it 'manually' in windows by doing a 'non-quick' format (full write scan) and a chkdsk X: /R scan (full read scan). This can be very time consuming but it is worth it. Best example "it was worth it" was spurious CRC errors on a batch of 6 brand new 4 TB drives (back when they were the bleeding edge capacity). If you return a drive with faults to a retailer within 7 days they will often bin it as DOA and refund or replace immediately.
Its nice to see some correct information on Chia farming (mining). With Chia farming, plots are written to the drive one time (maybe two ) and are very light weight. I would rather buy a chia farming HDD than a data center HDD or even a personal HDD as even personal HDD's are used more than Chia in a lot of cases
I am more concerned with the drive type than its use. Heavily used datacenter pulls have treated me far better than anything you can get from best buy.
@@andrewt.5567 Its just how hard drives work. If new one survived first two years - it will live for a decade or two. Spending these two years in datacenter is basically best reliability test you can do.
The irony of raid is you can increase your redundancy by adding more spares to your array while increasing your failure rate.
Sort of, you increase the likelihood of a drive failing, but you reduce the likelihood of your data being lost. (assuming you aren't using straight up striping with no redundancy)
I bought 10 3TB drives for about £150, 8 went into a nas as raid 10 (I wanted speed and redundancy) and I also have a new external 10TB drive used as a backup. This means I have 2 cold spares ready to go and saved alot of money on the initial cost with 0 drive failures in almost a year i consider it a win.
I have a several drives(HD103SJ) that are hitting between 70k and 85k hours. But they spin down when inactive? So not sure how much of that is actual active use on the hour count.
Sorry, but 2/3 of the price of a new one isn't "really cheap" for a hard drive
Even a new HDD is not worth risking your data. No substitute for backups. 3-2-1 and you’re probably good. I sleep at night with DVD-R and Amazon S3 deep archive REPLICAS.
Used hard drives are like used condoms: just not worth it and if it fails you're in for a nasty surprise
Well that's an interesting way to put it
Unless you are building large raid arrays where you have fault tolerance. For the price diff you can keep hot and cold spares around.
I do food reviews while I’m high off that good tree on my yöutube chånnel 😎
@@SevenHunnid hello spam
I bought a 4tb drive second hand 2 years ago and diskinfo said its basically new. no lost data, or hiccups over 2 years
Those prices for the 18 and 16 terabyte drives are always hella good 200 or $300 is amazing that's how much at one TB was when they first came out so that's pretty good amazing actually
An 18TB Seagate drive is $300 new at best buy right now, so it's not that great a deal
This video should have pointed that shucking drives usually means you end up with a slower 5200RPM drive and a worse warranty (or a voided one altogether) than a regular internal drive. You save some money but you take your chances.
My old shipping co-worker flipped some drives on their side on a table once, like this video (3:00). I marked them as "will fail," they were dead about a week later. NOT knowing how their treated makes them almost worthless.
Shoutout to the writers and Alex on this one, really well presented! The chronology and dialogue was well written and easy to follow. shoutout editor too and film crew too.
I prefer new, but have gotten used drives without much problem as well. Refurbs are generally as good as new in my experience, or at least close enough
Generally they just don't last quite as long, but depending on use case could last for years. Stuff for servers or previously in servers is a bit different tho
100% same here. All my hobby machines run on cheap refurbished 6€/250GB 2.5" HDDs and NO issues EVER! 💪
At 5:39 there appears to be an error in the chart, if your words and the following images are to be believed. Chart shows Seagate and Toshiba as having question marks (Unknowns), but verbally you said WD and Toshiba. A few seconds later, it looked like you showed a Seagate having given all the attributes.
If you go the external drive shucking route and pick up a WD, you'll likely need to cover the 3rd tooth of the power connector on the drive for it to work internally due to requesting extra voltage (as a measure against such conversion).
I had to do so with a 12TB ext I picked up.
This has nothing to with WD discouraging shucking. The SATA specifications changed in revision 3.3 (2016), they added support for Power Disable which allows you to remotely reset a drive. Usually only server drives have this feature but WD uses server drives for their external drives. Consumer PSU's haven't caught up spec wise to fix the issue so it continuously supplies power over that 3.3v pin causing the drive to be in a constant reset state.
If anything it's a problem with the PSU's.
I've had the most luck with Toshiba drives, got a few with bad sectors, one of which has ~3500 of them, yet there are no slowdowns, no clicking, nothing, it's like there's no issue at all. Same with the other Toshiba drives, altho less bad sectors.
On the other hand, WD and Seagate with bad sectors behave in many ways, from slowdowns, clicking, even grinding, some refusing to show up at all.
So for me, getting a used Toshiba drive has never been an issue, even ones with bad sectors. The other 2, i try to avoid used unless i can test them first.
If it's just for games, there's nothing wrong with using a used HDD, though of course there is always a good deal of risk that comes with buying one
Hard drives aren’t the best for gaming especially when they are used
@@rishiktandra7236 Of course they're not ideal, but they're not bad if you're on a budget
HDD for games... LOL ... have fun during loading.
@@PeterLunk i play cyberpunk- Witcher 3 -GTA 5-CSGO with an HDD, shit works fine and the loading aint that long
@@longerthanthirtycharacters I guess
Another good thing to note is the quality level of the used HDD. You may find consumer grade options that are the same TB/$ as the used enterprise grade, but the difference in quality and longevity can be worth it. Though, that can also come with its own set of caveats as some enterprise drives run much faster and more power hungry than consumer drives that are more likely to be idle. Used enterprise drives can be great though, especially if you go through all the SMART data like you guys showed! I personally have a combination of new and used enterprise HDDs (I have more storage at my house than the large OEM I work for...), and have had great success with both. It's always great to see this sort of information clearly conveyed so that people that might be on the fence will have a bit more confidence with tinkering around with used hardware!
Smart data can be reset easily with program. I have for WD. Bad blocks? gone. Power on hours, zero. Good as new. Don't trust smart data, do full scan, even on brand new drive, can be surface issue. Smart does not detect the errors unless you try to use each block. Hdd does not have a background integrity check.
I’m of two minds here. On one hand, I guess you could store games on used hard drives to minimize the risk. But on the other hand, you’d be much better off storing games on an SSD for the sake of load times.
Meh, with the size of today's games, (COD looking at you..) plus they are only getting larger, a 250gb SSD new for the price of a 2tb used HDD to me isn't worth it even with a few seconds extra loading time. Plus as it's only the game data stored if the drive fails you can just buy another one and reinstall the game. Furthermore, if you're not running the system of the hard drive and it's purely the game data only stored you will find loading times are not as bad as you would think.
Be careful when SHUCKING a hard drive, some western digital will act dead because they are getting power to one of the power pin on the drive where the external case doesn't provide power to that pin, you have to bend that pin until it breaks and then plug it in normally
big fan of the 1 1 1 rule lmao
"... a dedicated NAS has a singular purpose." - I want to correct Alex on that. Synology NAS as show in the show as a comparison to the home made NAS is NOT only a NAS! It can also be: Webserver, email server, VPN server, WebDAV server, Database server, Home Assistant, iTunes server, DHCP server, DNS server, LDAP server, Git server, SSO Server, Chat Server, Home Assistant server (via Docker) and more.
And THEN comes all the other standard file sharing features like, Antivirus, Clud Sync, Backup Vault, Media Server, Documents Server, Download Station, SMB and more....
The extra 722$ is counted for mainly all the hussle that one individual would have when he tries to install all these services and use them on its own home made NAS.
Unless somebodys pure goal is to have a dedicated external data vault with file transfer capabilities via SMB or other methods, then heck yeah! Best sollution ever and fully supporting it. But as soon as somebody would love to have some nice to have features like getting rid of Googles monopolistic cloud services and move all the data to it's own, then something like Synology is priceless because everything is at place. Just click and voila. If you even have a dedicated IP address, you don't even need the NASs provided quickconnect or any DynDNS.
So while I'm fully supporting the video and it is a great idea, the sentence "... a dedicated NAS has a singular purpose." triggers me a bit because there is the reason why it is 722$ cheaper to build you own NAS.
I bought used HDD most of the time... But I usually choose the youngest one, and choose the better one like the Red WD. I also asked the seller to test them with HDD sentinel / HD Tune too.
I just bought a lot of 9 4TB HGST SAS drives. They were like $15 per drive. They were manufactured in 2015, but had just under 40,000 hours on them (~4.5 years)
I'm putting them through their paces right now with badblocks under Linux. So far so good, no issues. All but one had very low, or 0 levels of ECC correction counts in the SMART data. None had uncorrectable error counts. (Since they are SAS, the data output is different than SMART on SATA drives)
I wouldn't use drives like these in a client, but in backed up redundant pools, why not? I just use more redundancy than I usually do. If one fails, I can get another $15 replacement. That beats spending thousands of dollars in drives when setting up a new pool.
I absolutely love HDDs and especially for my Hobby machines where no critical data is stored, these are fantastic 💪😎 In Germany there are always tons of used 2.5" HDDs with around 6€/250GB or 10€/500GB. All by companies with warranty. Can't go wrong.
Ssds are better
I have only ever had 2 HD's fail. The first one was in 2005 and 2 years old. Lost the first 3 years of my sons family photos, lesson learned. The IDE replacement is still going strong, but only used for backing up backup and just 120GB. The only other was my first SSD, which lasted 3 years, but even out of warranty, OCZ replaced it with a upgrade. This PC has two Cex specials that i have had for 5 years and still going strong. You can buy used with confidence so long as you backup your important stuff.🙂
Come to Australia, where a 16tb drive will cost you $730 Aud, God us Aussies get shafted on electronic prices