[HN Gopher] Chia Coin Miners Are Reselling Used SSDs as New
___________________________________________________________________
Chia Coin Miners Are Reselling Used SSDs as New
Author : thedday
Score : 137 points
Date : 2021-09-06 18:30 UTC (4 hours ago)
(HTM) web link (www.tomshardware.com)
(TXT) w3m dump (www.tomshardware.com)
| anigbrowl wrote:
| I'm sad about it, but I'm deliberately missing out on a lot of
| good crypto opportunities because I don't want to spend time
| sifting through all the fraudulent/scammy ones.
| withinboredom wrote:
| Reminds me of a shop that would buy used CDs, DVDs, etc, buff the
| crap out of them, print out new art, shrinkwrap them, then sell
| them as new (not almost new) on Amazon.
|
| They went out of business after they were blacklisted on
| Amazon... well sorta. They just came back as a different company
| a few times.
| lwansbrough wrote:
| Cool technology proof of waste is!
| sschueller wrote:
| I'm going to build a social network with a coin that works on
| proof of klout. /s
|
| In all seriousness it would be an interesting social
| experiment.
| tdeck wrote:
| Don't worry, apparently they'll be moving to proof of wealth
| soon which is Much Better.
| Dylan16807 wrote:
| Proof of waste is also proof of wealth but with extra steps
| and bigger environmental consequences.
| garmaine wrote:
| Isn't that exactly what proof of stake is?
| tdeck wrote:
| Yes that was my sarcastic point :).
| lvs wrote:
| So, to be clear about what's happening here, you only need an SSD
| for fast plotting in Chia. So it sounds like they are plotting to
| fill up HDDs, then reselling the SSDs to hedge against the
| possibility that the coin's value is not going to rise. It
| doesn't mean they're getting out of the game. It just means the
| network space has stabilized, and it's no longer
| necessary/valuable to plot any more space. But if the exchange
| rate rises, you can bet the network space rises too, requiring
| more resource-intensive plotting. That's why Chia is only more
| efficient than PoW when it isn't in demand.
| ahnick wrote:
| > That's why Chia is only more efficient than PoW when it isn't
| in demand.
|
| Actually that's incorrect. Chia is always more efficient than
| PoW, b/c in PoW everyone is constantly racing to solve every
| block. (i.e. find the proof) Whoever finds the proof gets the
| block reward and everyone else just throws away all that work
| for nothing and then the next race starts for the next block
| reward.
|
| Contrast that with Chia, where when you are plotting you are
| storing your proofs in a file for later challenges. You aren't
| discarding that work and you can hold onto it for a long time
| (years). It's not until you are "farming" the plot that you
| start answering challenges. The challenges just translate to
| simple reads from disk, which are quick constant time lookups.
| Now it may take you a long time (or never) to find a reward for
| a single proof, but at least you didn't have to throw all that
| work away and start over.
| lph wrote:
| It's almost as if any crypto scheme whose mining entails the
| massive waste of a resource is going to cause problems.
| giantrobot wrote:
| B..b..but disruption! Blockchain! Crypto!! To the moon!
|
| /s
|
| Crypto is like Eternal September where everyone wields flame
| throwers.
| jakedata wrote:
| The interesting part is watching people learn to become junior
| sysadmins - running Linux for the first time, teaching themselves
| a bit about storage, learning some shell scripting, dealing with
| firewall ports. Also learning the hard way not to run every scrap
| of PowerShell or Bash someone posts on the Internet.
| bluedino wrote:
| I'd rather they do this with Wordpress or Minecraft
| oh_sigh wrote:
| What's the punishment for this kind of sales fraud? Is it even
| possible to detect that the SMART counters have been reset?
| MertsA wrote:
| If you were suspicious that a drive was used before you ever
| plugged it in I bet you could identify it by immediately taking
| an out-of-the-box SMART snapshot and compare those values to a
| known good sample for that model drive. Drive manufacturers run
| tests when new, I doubt these Chia miners are trying to match
| up the powered on time and cycle count to a legitimate drive,
| they're probably just setting it to 0 and throwing it in a box.
|
| Unfortunately I highly doubt the majority of these fraudsters
| will face any real repercussions. Maybe a few victims will be
| able to successfully issue a chargeback, after many months
| maybe PayPal and other payment processors will try to freeze
| some funds, but whatever happens the fraudsters will still come
| out ahead. They'll just start selling under a new name and
| continue for another 6 months.
| gfosco wrote:
| Important distinction between chia drives used for plotting
| versus storage. The intensive write use is during plotting and
| would be on the smallest and fastest drives, in the 1-4TB range.
| Personal experience, I generated almost 400TB of plots and never
| had a drive go bad or experience any major performance issues.
| The massive storage arrays barely get touched, small reads only.
| lordnacho wrote:
| Don't SSDs store information about how often they've been
| written?
| Newtonip wrote:
| You are correct; there are even tools to view this information.
| Perhaps these "refurbishers" know how to reset the health data
| stored in the SSD.
| pkaye wrote:
| Yes, to view the stats you can use the SMART commands which
| is fairly standardized. But the ability is reset is usually
| some vendor specific commands. Its possible the tools leaked
| from manufacturing sites.
| Ekaros wrote:
| One would wish there was some burnable fuses there that would
| note this or general wear level...
| ectopod wrote:
| Yes, but I guess that a committed crook can clock drives just
| as they clock cars.
| rsync wrote:
| Word on the street is that they are resetting the SMART data
| with OEM firmware tools.
| a2tech wrote:
| The flash has never been touched right? Aren't they functionally
| 'new'?
| ceejayoz wrote:
| https://www.pcgamer.com/chia-mining-can-wreck-a-512gb-ssd-in...
|
| > Chia farming is a write-intensive activity. Speed matters, so
| the most common strategy is to use an SSD for creating plots,
| because SSDs are much faster than HDDs, and then transfer them
| to an HDD once completed.
|
| > Chia is a different animal, though. According to MyDrivers,
| mining Chia can trash a 512GB in 40 days, while a 1TB SSD lasts
| twice as long, and a 2TB SSD can give up the ghost in just 160
| days, or barely over five months.
| Ekaros wrote:
| Does that mean that bigger disk doesn't actually increase the
| speed which currency is generated?
| ceejayoz wrote:
| You'd be limited by write speed.
| qeternity wrote:
| A bigger disk means you can store more "plots", which are
| the the files necessary for proof-of-storage, and thus
| makes it more likely that you will earn a block reward..
| But in order to first generate these plots, you have to
| "farm" them which is a very write intensive process which
| is ultimately what is at focus here.
|
| The creation of these files requires so much writing that
| you are basically writing as fast as your CPU can, until
| the drive dies.
| ahnick wrote:
| The amount of XCH (Chia's currency) generated is a constant
| 64 Chia every 10 minutes. You can find this under the Chia
| Business Whitepaper (https://www.chia.net/assets/Chia-
| Business-Whitepaper-2021-02...) in the section "Post-launch
| Chia Emission Schedule". If you have a bigger disk that is
| filled with plots, then you are increasing the security of
| the network and thereby increasing the odds that you will
| win the block reward.
|
| For example, if you had 36EiB of space filled with plots
| then you would be equal to the entirety of the current
| Netspace of the Chia Network.
| (https://www.chiaexplorer.com/blockchain/blocks) So you
| would win the reward half the time and the rest of the
| network would win the reward the other half of the time.
| (i.e. 32 XCH every 10 minutes) Of course, this is an absurd
| example, as to obtain that amount of space it would
| probably cost you on the order of $20 billion just to
| procure the equipment, but at least it should give you an
| idea about how disk size works in regards to Chia.
| genewitch wrote:
| Let's say I have 1.5TB of ram, can I just do this in tmpfs?
| The reason this eats SSD is because they have to physically
| heat a chip to erase, right?
|
| Also, does anyone want to buy my server with space for 1.5TB
| of ram? The 425W of idle power usage became too much and I am
| moving to Ryzen 5950k
| humps wrote:
| Not according to this https://www.tomshardware.com/news/chia-
| crypto-farming-can-de...
| sithadmin wrote:
| The opposite of this. Chia is rather aggressive about issuing
| writes to disk and quickly chews through consumer grade SSDs'
| write endurance.
| zozbot234 wrote:
| There's a difference between creating new "plots" to farm and
| farming an existing plot. The former is very write intensive,
| the latter is a pure read workload.
| formerly_proven wrote:
| Well, yeah, that's why you use hard drives for storing
| plots (waiting for the distributed lottery to draw your
| plot's number) while continuously generating new lottery
| tickets on the SSDs.
| jazzyjackson wrote:
| Still, one of the coin's selling points is that low-energy
| for farming and verifying transactions. That is kind of
| moot if there's an economic incentive to burn SSDs to get
| your hands on these eco-friendly coins.
| mangecoeur wrote:
| For something that was supposed to be more "green" this seems
| like a very obvious oversight
| Slartie wrote:
| If you want some new crypto to succeed, it must have at
| least one blatant lie on the first page of its white paper.
| Otherwise no "investor" will buy the stuff.
| Dylan16807 wrote:
| Yeah, the plotting thing is really bad.
|
| But thankfully: 1. the network has mostly stopped expanding
| 2. I think there's a pretty good ramdisk plotter now for
| big miners
| ceejayoz wrote:
| https://en.wikipedia.org/wiki/Greenwashing
|
| "Dishonesty? In my crypto? It's more likely than you
| think."
| rapind wrote:
| Reminds me of the argument this guy was making recently
| https://www.cbc.ca/news/canada/calgary/link-global-
| bitcoin-m...
|
| > "We look at, OK, what can we do to use this in a
| beneficial way ... I don't want to say we're in the
| business of methane destruction, but we're in the
| business of beneficial use of that potential methane-
| generating source. You combust it properly. You don't
| flare it, and you control those emissions," Jenkins said.
| gruez wrote:
| The article isn't exactly clear what's happening. Are
| they extracting natural gas (methane) that would normally
| stay in the ground, or are they flaring natural gas that
| would otherwise have leaked out?
| wmf wrote:
| This should be obvious, but anyone can sell any used hardware and
| claim it's new.
| jazzyjackson wrote:
| The chia angle is quite fascinating tho, it is a machine that
| sucks the lifespan out of memory chips and cashes out with
| crypto-tokens. It is an invention that gives economic incentive
| to burn as many memory-cycles as you can before reselling the
| SSD, such that the recipient of the "new" SSD is ripped off.
|
| It's kind of like how the electricity to create a bitcoin costs
| more than the market value of the mining reward, the only way
| to be profitable is to steal electricity. In chia's case, the
| best way to profit is to steal memory write-cycles.
| londons_explore wrote:
| Compared to database-in-datacenter type uses, chia actually
| puts very little load on ssd's.
|
| All writes are fully streaming, there is no write
| amplification, and there are lots of random reads involved
| (which don't really cause much wear/stress).
| jazzyjackson wrote:
| There is still a finite capacity, say 1,000 Terabytes
| written to a 1TB drive. You can do it as gently as you
| want, it is still incentivizing the writing and overwriting
| of exabytes of noise.
| imglorp wrote:
| I think the subtext is, "yet another scam perpetuated by the
| cryptocurrency community at large."
|
| The whole thing started out as a utopian ideal for freeing
| people from centralized control of transactions and it's
| turning out to be yet another thing ruined by bad actors.
|
| In that context, buying SSD's from (largely) thieves who
| (largely) will fleece you in several other ways will not turn
| out well for you.
| agumonkey wrote:
| there was a thread on HN about 'boring advices' and every
| 'utopian ideals' seems to be completely out of that class :)
| nyolfen wrote:
| >I think the subtext is, "yet another scam perpetuated by the
| cryptocurrency community at large."
|
| sounds like it's specifically the chinese crypto community
| rsync wrote:
| This is very relevant to our interests - specifically the part
| where prices of spinning disks _more than doubled_ for a time
| earlier this year.
|
| As we dove into the chia coin rabbit hole to figure out just what
| was going on, it came to our attention that people are using
| firmware tools to reset SMART counters on drives ... rolling back
| the odometer, so to speak.
|
| Just terrible behavior all around.
| ksec wrote:
| HDD price hasn't been lowered for quite some time due to
| technological bottleneck. Even with OptiNAND only give us the
| same density as SMR. And EAMR are no where to be seen or tested
| by consumers.
|
| And now this.
| londons_explore wrote:
| I bought an SSD on eBay recently that massively underperformed
| my expectations.
|
| But I have no real proof to send it back for a refund. Instead
| I'll just have to suffer with super slow writes.
| Cacti wrote:
| 1. verify it against published benchmarks for that model 2.
| open up the case and see if it's actually what you bought
| kkielhofner wrote:
| In my experience the eBay/Paypal resolution process is so
| heavily weighted to the buyer you shouldn't have any issues
| returning it and/or getting a refund.
| dheera wrote:
| Also, if someone is selling _anything_ on eBay for above
| MSRP I report them for price gouging. Masks, sanitizer,
| drives, Ubiquiti cameras, anything. All reported.
| avalys wrote:
| What will you do if the manufacturer increases the MSRP
| because demand exceeds supply? Report them for price
| gouging?
| dheera wrote:
| If it's an emergency situation they shouldn't be
| attempting to profit off the emergency, and it's illegal,
| immoral, and an asshole move.
|
| If it's not an emergency situation then I'm okay with
| them increasing MSRP. They're the ones who make it, they
| set the price.
| AussieWog93 wrote:
| This. I sell an eBay for a living and unless I have proof
| beyond reasonable doubt that the buyer is lying, I am
| required to provide a refund.
|
| Since I started telling buyers that I log serial numbers
| for every item I sell, though, a surprising number of them
| started closing their own cases.
| bserge wrote:
| Just great. I mean, drives are the only things I'd rather buy
| new, but how hard can it be to fake the whole packaging these
| days.
| ev1 wrote:
| Miners frequently retain the full packaging for return
| fraud/resale purposes.
| throwaway9980 wrote:
| Dumpster divers are also recovering packaging for re-use.
| 55873445216111 wrote:
| Buy direct from manufacturer. Samsung, Western Digital, etc
| all have their own online stores.
| Nexxxeh wrote:
| Having purchased an "in stock" SSD directly from WD at the
| end of last year, it's not something I'm keen to repeat.
|
| Their store seemed to operate primarily on incompetence.
|
| I finally got my invoice two weeks later, still with no
| working shipping information, and the drive turned up days
| later still.
|
| On top of their scummy behaviour around SMR disks, and
| their NAS solution, I think I'll be steering as clear as
| possible.
| bogwog wrote:
| And sidestep the monopolist at the same time. Its a win-
| win!
| ahnick wrote:
| What people who read this article may not realize is that this is
| the Chia incentive model actually working properly. It is not
| cost effective to buy new hard drive equipment to farm(mine)
| Chia. (This was well discussed by Chia Network and the community
| before mainnet launch) Chia is really designed to be used to farm
| existing available storage space that people or companies already
| have for other purposes. For example, a home NAS system that may
| currently only be using a small portion of it's capacity. People
| can then slowly fill that space overtime and potentially make
| some money from their unused space.
|
| The only time it made sense to buy new equipment to try to plot
| really quickly, was right at the start before Netspace increased.
| Chia now has somewhere on the order of 300,000 - 400,000 full
| nodes operating on the network (already making it more
| decentralized than Bitcoin) and is approaching 36EiB of space.
| The "green" benefit of Chia is that you don't throw away the
| proofs that get generated like you do with Bitcoin mining and
| instead you can keep them for many years and potentially decades.
| This is where a lot of the power savings come from. Take a look
| at https://chiapower.org/ to understand the power usage compared
| to existing cryptocurrency networks like Bitcoin and Ethereum.
| Krisjohn wrote:
| No, it's not.
|
| Chia is just "proof of prior work" and as such has the same
| arms race problem as bitcoin. People aren't just buying a few
| TB, farming it and being content. They're building farming rigs
| and piling on the storage constantly. It just consumes power
| AND storage.
| ahnick wrote:
| Again, it was only cost-effective to build farming rigs in
| the very beginning. It is no longer cost-effective to do so.
| Even major companies will eventually have a tough time
| competing with the amount of storage space that is already
| deployed in the world. IIRC is around ~7 Zettabytes. That is
| the big difference with Bitcoin. Chia is making use of a
| resource that already exists in the world and is well
| distributed. Bitcoin ASICs require manufacturing and building
| machines for the specific purpose of mining Bitcoin and then
| distributing those machines.
| matheusmoreira wrote:
| What matters is whether people are churning through hardware,
| not the scale of the mining operation. Apparently this churn
| was a problem in the past but it's not anymore. Is this not
| correct?
| nathanvanfleet wrote:
| I definitely want to destroy my Home Nas in a matter of months
| to make a little money. This makes perfect sense. It seems like
| this is the energy problems of bitcoin but now it's physical
| hardware that is the inefficiency? I guess Bitcoin is similar
| with old miners becoming defunct etc and the cost of video
| cards. But intentionally running down the life of hard drives
| is a innovative new feature I guess
| ahnick wrote:
| > I definitely want to destroy my Home Nas in a matter of
| months to make a little money.
|
| That won't happen unless your NAS is a bunch of consumer
| grade SSDs, which I hope it is not.
| X6S1x6Okd1st wrote:
| The video cards were being used for ETH and altcoins.
| Thankfully ETH next upgrade is to turn off proof of work.
| nyolfen wrote:
| > or example, a home NAS system that may currently only be
| using a small portion of it's capacity. People can then slowly
| fill that space overtime and potentially make some money from
| their unused space.
|
| using my spare disks like this does not make sense when it
| apparently burns them out very quickly. i doubt i would recoup
| even the cost of a new drive
| ahnick wrote:
| This is only really true when plotting to consumer grade
| SSDs, which are limited in the number of writes that they can
| support. NAS systems usually contain mechanical HDDs and can
| handle the write load of Chia without issue.
| qeternity wrote:
| I think this was missed in this whole issue. I'm not
| arguing for or against Chia, but basically in order to
| bootstrap a mining operation faster, miners were generating
| the initial files on a much faster SSD and then
| transferring to HDD. The initial bootstrap ("plotting") is
| very write intensive, and thus destroys the SSD. HDD do not
| have the same sensitivity to write heavy workloads.
|
| So if one simply plotted very slowly on an HDD with the
| spare space they had, it would not likely impact the
| expected lifespan, and they could earn a few pennies for
| their efforts. However the issue with all of these crypto
| efforts is that systems naturally trend towards
| centralization for reasons of efficiency. The best systems
| have some balance between the two. It's simply not worth it
| for most people to plot 100gb of free space on their laptop
| for Chia and earn a few cents per month. So it only makes
| sense for miners to do at scale.
| ahnick wrote:
| Yeah plotting on a laptop with a 100GB free is not worth
| it (especially since they typically have consumer ssds in
| them), but there is a lot of spare storage space on
| servers/external drives owned by individuals or companies
| globally speaking. This global storage distribution that
| already exists and the fact that no specialized hardware
| (ASICs) are needed is why Chia has been able to become
| more decentralized than any other cryptocurrency
| currently.
|
| It will be interesting to see how things centralize or
| decentralize going forward. When I think of the lottery
| I'm always amazed by the number of people that play it
| even with the extraordinarily low odds of winning that
| exist. I think there will be some equilibrium that's
| found for the global number of operating nodes that will
| be a function on the odds of finding a proof and Chia's
| price.
| nathanvanfleet wrote:
| That's not how hard drives work
| skybrian wrote:
| Yes, it's probably too late. Or so I've heard.
|
| Chia is like Bingo. Generating the Bingo cards can put a lot
| of wear on an SSD, but once you have the bingo cards to fill
| up whatever space you have, you can play for nearly free.
|
| I haven't done the calculation myself, but I've read that
| it's no longer worthwhile to generate new cards. If so, that
| means most of the damage has already been done. It's a sunk
| cost.
|
| But the people who are already in the game can keep playing
| indefinitely. If prices are too low to even do that, they
| could back up to tape and go offline for a while.
| twalla wrote:
| Plotting, or generating proofs, is what destroys SSDs,
| afterwards the plots are stored on disk and aren't read/write
| intensive. Think of it as printing bingo cards (plotting) and
| then sticking them in a file cabinet (farming). Your printer
| consumes ink and electricity, but keeping all your printed
| off bingo cards only costs you the space the file cabinet
| takes up and maybe the lights you turn on to look for winning
| cards when they get called.
| SrslyJosh wrote:
| > What people who read this article may not realize is that
| this is the Chia incentive model actually working properly.
|
| I think that most people commenting here understand that this
| is the way that proof-of-* systems end up functioning at scale.
| Participants in these systems are incentivized to cut corners,
| externalize costs, and generally fuck things up for everyone
| else.
| tus89 wrote:
| There is a special place in hell for these cryptards.
| moogly wrote:
| Amazon Marketplace?
| tus89 wrote:
| Too far.
| bserge wrote:
| Worse, there is no hell.
___________________________________________________________________
(page generated 2021-09-06 23:01 UTC)