Hacker Newsnew | past | comments | ask | show | jobs | submit | metadat's commentslogin

Seagate continues the tradition of having the highest failure rates of any manufacturer, on average.

Why is that?


I do not know, but the last time when I have bought a Seagate HDD, I had a very nasty and unpleasant surprise.

Last year I have bought a 22 TB Seagate Expansion Desktop external HDD, because it was cheaper than the other 22 or 24 TB HDDs available at that time.

I had read carefully its datasheet before buying and there was nothing suspicious there, so I assumed that it must be cheaper just because it is a slow HDD. I did not care about the speed, it was for storing data archives infrequently accessed.

Only after receiving it I discovered what was not said in the datasheet, that this Seagate HDD does not support S.M.A.R.T., so there is no way to test it to see if it works OK and there is no way to discover when errors have happened, e.g. to see when the HDD becomes too old, so you need to migrate your data.

I have never imagined that in 2025 it is possible to buy a HDD that does not support S.M.A.R.T., especially in HDDs with a capacity over 20 TB, and moreover without giving a prominent notice about such a misfeature in the datasheet.

Before this, in 2024 I had bought a 24 TB Seagate SkyHawk, which had S.M.A.R.T., as expected. Since then, after the Seagate Expansion fiasco, I have bought a 22 TB external WD HDD, at the same price with the Seagate, and which has S.M.A.R.T., as it is normal.

I cannot see how removing S.M.A.R.T. support can reduce costs, as it is just a firmware feature. I any case a manufacturer that removes testing and error reporting features from its products clearly does not give a s*t about data corruption and HDD failure rates.


Doesnt surprise me, Seagate is marching to its own drum. My experience defiantly mirrors others' higher than average failure rate as well.

My latest 'fun' experience with them, also, came in the form of an Ironwolf drive which is 'detected' on usb-to-sata interface when plugged in, around %15 of the time. While it starts up consistently on a plain SATA interface. This makes it unusable for what I need. Again, no other drive or MFG ever fails on this usbSata, just the new ironwolf, which it appears is actually for the chineese market, but was sold on newegg, but this is not necessarily seagate's fault, nevertheless.


With an external drive the SMART info might be hidden behind the USB-to-SATA bridge, smartctl has support for some of those but sometimes needs to be told with an extra argument.

Do you have MacOS by any chance?

MacOS does not support S.M.A.R.T over USB.


>there is no way to discover when errors have happened

There is: use ZFS and scrub.

But yeah, crazy that it doesn't support SMART!


I have two particularly notorious Seagate periods:

    Seagate bought Conner when Conner had released several models w/ 
    leaky seals. Bad sectors started at the outer edge of the 
    platters and grew inward. We had a lot of these drives
    out there and Seagate refused to honor Conner's drive
    warranties. 

    The 7200.10 series had super high failure rates. I wound up 
    replacing every one in my care, within 2 years. The 7200.11
    drives weren't much better.
I think the last Seagate lines I truly trusted were the ST series of MFM and RLL drives.

As explained at https://www.backblaze.com/blog/backblaze-drive-stats-for-q3-..., a large proportion of Backblaze's Seagate inventory are rather old drives for a datacenter (now 5-9 years in service), so a high failure rate is expected.

I have quantum fireball from 2000, so 26 years old, still going strong.

5 years doesn't seem that long for a drive that cost hundreds of dollars! Persistence is the point.

Just wondering why Seagate seems like the bottom of the barrel in the longevity department. Western Digital drives seem to fail a lot less frequently on average in this dataset and in my life experience.

To Seagate's credit, I do have 8x24TB drives that have been working fine for the past 4 years. Hopefully can last a few more until the compute hardware shortages pass.


That's a brand I haven't heard of in a long time. I had a 8gb HDD from the brand in 2000 until my sister kicked the computer case out of frustration which ended up shorting some chips on the drive. I mourned the loss of my music collection for quite a long time.

I used a 1.2gb Fireball for my main drive around that era — it is so comically loud!.

Well, I have a 200mb maxtor IDE works just fine to this day.

The classic Deathstars™ ["Deskstar"]

Only expected if its Seagate. Backblaze Hitachi drives had miniscule failure rates thru their whole life cycle.

It actually looks like they're getting better, if the changes from last year to this year are any indication.

"back in my day", seagate was "the shit". only much later, hitachi drives came to be popular and wd, sort of.

https://factory.strongdm.ai has some advice I've found useful.

That is a useful pattern, though I was unclear on why `t.Cleanup` and not `defer`. In case others are curious, too:

> Parallel subtestsWith t.Run(..., func(t testing.T) { t.Parallel(); ... }), the parent test function can return (and thus run its defers) before parallel subtests actually finish.*


Short version is this:

If you are going to get into the business of introducing order dependence to test cases through global state (see my other reply on the parent), you will always want the cleanup to work correctly.

1. Using (testing.TB).Cleanup is a good defensive habit to have if you author test helpers, especially if the test helpers (see: (testing.TB).Helper) themselves do something (e.g., resource provisioning) that requires ordered teardown. Using (testing.TB).Cleanup is better than returning a cancellation or cleanup function from them.

2. (testing.TB).Cleanup has stronger guarantees about when it is called, especially when the test case itself crashes. Example: https://go.dev/play/p/a3j6O9RK_OK.

I am certain that I am forgetting another edge case or two here.

Generally nobody should be designing their APIs to be testable through mutable global state. That solves half the problem here.


What's your email address?

Just get the damn RSS

The irony of stealing product activation is WOW :) welcome to capitalism.

welcome to humanity: do as I say, not as I do

How does it count so fast? Clickhaus preloaded dataset?

Top 0.023%, I was surprised! I usually keep it pretty short here, and my account isn't old.


How is attending a protest a potential violation of customs regulations? This doesn't track.


The argument, if there is one, would probably be that following ICE was a violation of an immigration procedure (note that the person who had their GE revoked doesn’t claim they attended a protest, but rather that they were following ICE and got their picture taken). Given what I’ve seen of GE revocations historically, though, it’s equally likely to have been something like “you lived with a felon” or “unpaid traffic ticket became a warrant” or “family member was accused but never convicted of an obscure crime.”

There’s always been a pretty clear mantra that GE is a privilege not a right and that it’s always been an arbitrary and capricious system.

In some ways I think maintaining GE is probably as hard or harder than maintaining a low level (ie Secret) security clearance; it seems to be based on similar databases and discretion with less transparency, human touch, or opportunity to appeal.

They are at least (according to the 9th circuit) supposed to disclose why the GE was revoked though: https://cdn.ca9.uscourts.gov/datastore/opinions/2024/05/22/2...


> How is attending a protest a potential violation of customs regulations?

FTFA: “Protesting isn’t a listed or ‘valid’ reason for having Global Entry revoked, but being arrested at a protest is. Impeding or interfering with the agency is. And being investigated is.”


Looks like this is a social credit system.


Global Entry absolutely is a completely voluntary social credit system? It’s an optional thing you literally subscribe to that says “I’m squeaky clean so I can skip the investigation line.”


right, but as the post above says:

> “Protesting isn’t a listed or ‘valid’ reason for having Global Entry revoked, but being arrested at a protest is… And being investigated is.”

its no secret they're declaring everything a protest. they are actively and seemingly randomly roaming, demanding ids, and then declaring everything from that moment forward a protest. this isnt imaginary, its happening regularly.

it’s no secret how many people they have arrested and released an hour or two later once they’re "investigated."

it’s also no secret that they’re simply arresting people, who arent at a protest, but who happen to be in the same street they’ve roamed to.

if they roam into my neighborhood, and if i were to be outside walking my dog, they demand my id which i probably wouldnt have because im just walking my dog. so they arrest me. hours later let me go because they’ve "investigated" and oops, just a neighbor out walking his dog..

if they've declared it a protest, this would mean i would have just been arrested and investigated at what they call a protest.

so now in this situation i would barred from global entry because they’ve declared everyone in their eyeline worthy of arrest and then wrongly arrested and investigated me?

what a cluster fuck. we need to get due process back, this is insanity.

not to mention that protesting, just like legally carrying a firearm is absolutely protected by the constitution and not at all a criminal offense. getting arrested at a protest because they dont like protesters is absolutely not a reliable indicator of any kind of illegality.


> Global Entry absolutely is a completely voluntary social credit system?

They’re making analogy to China’s social-credit system. It would be like if Global Entry was held by most Americans and you needed it to get a credit card or board a train.


> Looks like this is a social credit system

To the extent it’s a government program with any discretion, yes. In every other respect, no.


You appear to have successfully formed the habit of posting on Hacker News frequently, isn't this a starting place? :)


Touché.

Maybe instead of fighting it, I should build on it. Thanks!


Fine-tuning coder models is not nearly as effective as intelligently managing the context with frontier models (opus, gpt-5.2-codex).


I don't think it's even a question. A 32b model will not compete with SotA for years to come (if ever). The idea behind this release is to fine-tune on your codebase and compare to non-finetuned open models from the same class (or one higher). So if you need local processing, without access to SotA (security, compliance, whatever) then this is an interesting avenue for you. And the cost is fairly low. They are releasing the method to do this on your own codebase / docs / processes.


2.7% overall inflation for 2025 yields a $96B loss for insurers, wow.

Haven't they already been raking us over the coals enough with 10x health bills to absorb this without missing a beat?


That past taking has been amortized over many different VC fund losses already. Where did you think the people who invested into WeWork got the billions from? It didn't go into anything like due diligence.


You’re assuming there will ever be enough for these bastards.


It is a 96b loss in market cap, not revenue or profits.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: