Hacker News Reader

Comment by wink | Hacker News Reader

Maybe I'm the odd one out, but apart from the new cloud stuff 've had nothing but bad experiences.

Many years ago: HDDs died like I've never seen it before, one every few months with just a couple of servers.

ca. late 2013: We had some production systems running on Hetzner hardware, we regularly! called their support to tell them the DC lost connection because our monitoring apparently was better than theirs and they acted surprised for not hearing something yet.

Only like 2-3 years ago: Got a new server, had SMART issues and had to have the disk replaced before the first 24h after ordering were up.

An ex-coworker of mine joked they must have added the "please change the disk" option in the support form just for him because he was running 20+ servers and actually had to call them every week to replace a dying disk.

Yes, this was all years ago - but I'd rather pay servers out of my own pocket than be oncall for production servers there, ever again.

View on HN

Most of your comments are about disks. If you read the fine print, when you order a new server, you don't get brand new disks, but used ones. When they die, you don't get new disks, unless you pay a premium, but used ones. They recycle the same broken hardware until it's untenable, and then some more.

Didn't know this. Recycling is good for ecology.

At work we have some Hetzner servers. About two times a disk broke; with RAID it was not a problem, a hassle, at least the first time, it was, yes.

> Recycling is good for ecology

Looking at the recycling issue I wonder if you are using so many disks (that fail) if it is possible to ship them to a low wage (sorry) country have them taken apart and rebuilt using parts from other disks where different parts failed?

It's possible to combine the parts from a disk with failed bearings and a disk with a failed controller to end up with one working disk, but it's not worth the effort unless your goal is data recovery.

Hah why I said 'low cost country'.

Some vendors do that (look at recertified disks) but the failure rates aren't great.

All the dedicated server providers reuse components, it's normal practice.

As long as the health of the HDD is 100% with no bad sectors there's no reason not to reuse it.

When you get your new hetzner server check it with smartctl, hdsentinel etc, if any show bad sectors or health other than 100% I am sure they'll be happy to replace it for you.

I have had both disks in my Hetzner server replaced (at different times), but they are in software raid1 so hardly caused any downtime for me. They were extremely quick, efficient, and no-nonsense about it.

> I'm the odd one out

You are (from my 10+ year experience with Hetzner at least).

loving this comment: a: am I the odd one? b: based on my 1 datapoint yes (ok it's 10 years but still one customer?)

Your right of course, but as we don't know the Hetzner statistics it's hard to say something definitive. -- As you (wink/scotu?) provide 1 datapoint I was thinking 1 datapoint from my side should be defensible :-)

So, it seems my chance, if I go with them, is 50-50.

> many years ago: HDDs died like I've never seen it before

i ve also had some disk problems 5+ years ago, but not recently, even with the cheap bidding servers. Nothing that wasn't justified by their way-cheaper-than-others prices

This is exactly the reason why I switched away from Hetzner and I wouldn't switch back even if they were free.

I had a production server with two HDDs in some form of redundant raid configuration and both died at the same time. You somewhat lose the benefits of both drives mirroring the same data if both fail!

Hetzner is awesome. Been with them fro 5 years.. they're great!

I've used them for years. for the price I can't complain. usually takes 45 minutes max for a disk swap. the only part that can suck is dealing with abuse complaints, they're pretty heavy handed, but still fair.

Well, with all respect, you got what you paid for. This is the difference in between bare bones dedicated server hosting and managed hosting of any kind.

At one of my previous workplaces, we had a park of around 3k servers, with 2k of them on Hetzner. At such scale, you can not physically escape from the fact that all kinds of weird stuff happens with hardware, like buggy raid cards, physically unrebootable servers, dead cores, faulty and fake memory, bitrot, throttling due to underpowered PSUs,etc.

My favourite find was a "xeon 2637v3" with 6 cores (2 extra cores for free! unlike the normal sku.)

And we always had argument about their service, availability, giving us three faulty raid cards in a row, sued them few times for a refund, but... the company still liked them sooo much.

That doesn't make sense. You can still get barebones hosting and not get dying disks and bad support.

I don't know, how is the formula for "you get what you pay for"? Double the price = half the issues? Still not in Hetzner's favor. I've paid twice the price at other German hosters and had zero problems over many years.

Also I wasn't really complaining I'd like to think - I was just stating facts. I saw like 10 "this is so awesome" replies here, am I not allowed to say I've not been a happy customer?

Reply on news.ycombinator.com