Hacker News Reader

Comment by fortran77 | Hacker News Reader

You can build a little hot-swappable NAS with nice trays to slide disks in and out, an easy web GUI, front panel status lights, support for applications like surveillance cameras, etc, with junk you mostly already owned?

View on HN

I don't think most people consider easy hot-swaps + front panel status lights particularly key features in their home NAS.

I don't swap drives unless something is failing or I'm upgrading - both of which are a once every few years or longer thing, and 15min of planned downtime to swap doesn't really matter for most Home or even SMB usage.

-----

As for the rest, TrueNAS gets me ZFS, a decent GUI for the basics, the ability to add in most other things I'd want to do with it without a ton of hassle, and will generally run on whatever I've got lying around for PC hardware from the past 5-10 years.

It's hard to directly compare non-identical products.

For me and my personal basic usage - yes, it really was pretty much as easy as a Synology to set up.

It's entirely possible that whatever you want to do with it is a lot of work on something like TrueNAS vs easy on a Synology, I'm not going to say that's the case for everything.

In fact I find the synology disk trays to be very fragile. Out of the 48 trays I have, I think a good 6 or 7 do not close anymore unless you lock them with a key. A common problem apparently.

Hot swap for drives is a must on a NAS. If you have to power it down to swap out a drive there is a chance that your small problem becomes a larger one. Better to replace the drive immediately and have the NAS do the rebuild without a powercycle.

If you're worried the hard drives won't spin back up, I'd say you should instead spin them down regularly so you know that risk is basically zero. If you're worried the power supply will explode and surge into the drives when you turn it on, you should not be using that power supply at all. Any other risks to powering it down?

And for the particular issue of replacing a failed drive and not wanting to open up the case while it's powered, you can get a single drive USB enclosure to "hot swap" for $20. And if you use hard drives you should already have one of those laying around, imo.

Agree, you should consider replacing your drives on your primary server (backup servers we can debate) as soon as you start seeing the first SMART problems, like bad sectors. If you do regular data scrubbing, and none of these problems show up on the other drives, I'd argue the risk that they fail simultaneously is fairly low.

Hot swap drives are necessary on data centers where you don't want to have to pull the whole server and open the top cover just to replace a disk.

But on a home NAS? What problem would having to power it down and power it on for drive replacement create? You're going to resync the array anyways.

I don't mind them and I do use them but I consider them a very small QOL improvement. I don't really replace my disks all that often. And now that you can get 30TB enterprise samsung SSDs for 2k, two of those babies in raid 1 + an optane cache gives you extremely fast and reliable storage in a very small footprint.

> If you have to power it down to swap out a drive there is a chance that your small problem becomes a larger one.

What are you thinking of, here? Just a scary feeling?

No, I've seen this happen on larger arrays. The restart with a degraded array risks another drive not coming up and then you are on very thin ice. Powercycles are usually benign but they don't have to be and on an array there is a fair chance that all of the drives are equally old and if one dies there may be another that is marginal but still working. Statistically unlikely but I have actually seen this in practice so I'm a bit wary of it. The larger the array the bigger the chance. This + the risk of controller failure is why my backup box is using software RAID 6. It definitely isn't the fastest but it has the lowest chance of ever losing the whole thing. I've seen a hardware raid controller fail as well and that was a real problem. For one it was next to impossible to find a replacement and for another when the replacement finally arrived it would not recognize the drives.

Sure. You buy a chinese case with 6-8 bays off Aliexpress, throw some board with ECC RAM support into it and a few disks. You install TrueNAS Scale on it, setup a OpenZFS pool. Front panel lights are controllable via Kernel [0], it even offers a ready-made disk-activity module if you want to hack. Surveillance cameras are handled by Frigate, an open source NVR Software which works really well.

Especially when you want to build and learn, there's next to no reason to buy a Synology.

[0] https://www.kernel.org/doc/html/v5.3/leds/leds-class.html

Very valid advice, but you don't do all that in "an hour," of course. Synology's purpose in life is to provide a solution to users who are more interested in the verbs than the nouns.

They are the Apple of the NAS industry, a role that has worked out really well for Apple as well as for most of their users. The difference is, for all their rent-seeking walled-garden paternalism, Apple doesn't try to lock people out of installing their own hard drives.

Kudos to Synology for walking back a seriously-stupid move.

Once you have the case, an hour or two is pretty reasonable... you can even have your boot device pre-imaged while waiting on the case to get delivered.

Not to mention the alternative brands that allow you to run your own software... I've got a 4-bay TerraMaster (F-424 Pro) as a backup NAS. I don't plan on buying another Synology product.

I'm no stranger to building boxes or running servers, but I've run a couple of different Synology NAS over the past 15 years. My estimate is that if I were to put together my own system, it would probably take several days and cost about the same as if I were to buy Synology. I'm not familiar with building NAS systems specifically, so that might be part of the issue. But saying you can do it in one hour seems like hyperbole.

When I looked into it last, I planned to spend about as much as a Synology, but it would have much more compute, memory and as much storage. I was likely going to run ProxMox as a primary OS, and pass the SATA controller(s) to a TruNAS Scale VM... Alternatively, just run everything in containers under TruNAS directly.

For my backup NAS, I wound up going with a TerraMaster box and loading TruNAS Scale on it.

Someone building their own probably isn't too afraid of missing out on a webgui or installing something like FreeNAS or whatever is the popular choice these days.

I think the NAS market is in for an upheaval due to the markups for fairly crappy hardware and then squeezed from the bottom by cloud storages.

RPI 5 can be got with 16gb of memory and has a PCI-E port, some might complain about the lack of ECC ram but does all those cheap ARM cpu's on lower end NAS'es really have that?

I think the biggest factor might be that case manufacturers haven't found it to be a high enough margin, but it only takes one to decide that they want to take a bite out of the enthusiast NAS market.

Well, one man's junk is another man's treasure.

In any case, none of the requirements you listed seem that exotic. There are computer cases with hot-swap ready drive cages, and status lights (or even LCDs) are easy to find. The software is probably already on github. The toughest ask is probably for it to be "little", but that's not something everybody cares about. So I don't find the GP's claim to be that much of a stretch.

they’re pretty clearly referring to _their_ use case and not everyone’s. i think people are mostly talking past each other about this. there isn’t one feature set that matters for everyone, so of course a synology is perfect for some and for others it can be replaced with “junk”.

There are several drive tray cases for ITX and mATX that you can choose from. As for a Web GUI, you can get TruNAS Scale running relatively easily and there are other friendly options as well... so yes.

... and "it only takes an hour?"

LOL, clearly an amateur. That's longer than it took me to build Dropbox. /s

Reply on news.ycombinator.com