TrueNAS on a single Disk for a small home server?

Waste of what? For the sake of capacity, a SATA port for plugging another HDD is more valuable than a M.2 slot.

1 Like

Yeah but even a single NVMe drive for apps, backed up nightly to HDD, is an option. And NVMe drives are about about equal to an SSD in terms of availability and cost.

When I built my system in 2016, a Samsung SM951 128 GB M.2-2280 PCIe 3.0 X4 NVME Solid State Drive cost me $100, but I wanted to use all eight SATA slots for hard drives. Now, I can pick up a Inland TN320 256GB SSD NVMe PCIe Gen 3.0x4 M.2 from Microcenter for $24, or a RiDATA E801 128GB NVMe M.2 PCIe Gen3x4 from Newegg for $16. And, those are the smallest sizes and the cheapest that I could quickly find.

1 Like

Same, but i have only 6 :melting_face:
For boot, also “pseudo” 16gb Optane are very easy to find under 4€. At this point every motherboard Is capable to boot from PCI-EX slot using a cheap 2~3€ adapter.
Despite Is not optimal, in my experience i can say that also a 1x slot do the job reliably (i keep the faster slot for nvme dedicated to apps)

1 Like

Well, I typically reserve NVMe slot for VM’s; something that can actually utilize that speed.

Of course, everyone is free to use that slot as they see fit. I just see it as a lot of wasted potential to use it for the boot volume.

The quoted text said “TrueNAS software”, which is true, it was not designed to use combined boot & data volumes. When iX took over FreeNAS, they continued development using separate boot device & data device(s).

I am not trying to stop anyone from doing anything they want. My intent was to help people understand that building a minimal computer to use as a server might have odd performance or failure conditions not generally experienced by others using server grade hardware.


As an example, while ECC RAM is not required by ZFS or TrueNAS, we have seen some odd failures where a ZFS pool can't be imported due to severe corruption. Some pool corruption problems were tracked back to other causes.

However, a few never had a good explanation. My “guess”, (and it really is just a guess), is that some of those unexplained pool corruption cases were caused by memory faults or bit flips. Rare, but you don’t want to be the one that has such.


Another example is that people persist in considering power failures as ways ZFS pools get corrupt. This is not true, and I wrote about why that is not the case:

Of course, if the power failure causes a hardware failure, like damaging a disk, well, of course that could lead to pool corruption.


So, my intend is to pass information that might help a person make a reliable NAS and avoid data loss. But, they are free to ignore me, or discuss details
1 Like

Because that paper doesn’t address TrueNAS (or FreeNAS) in the slightest? Also note where ZFS was developed–Sun weren’t making Chromebooks; they were making high-end systems.

Rather than looking at the ZFS documents, you’d look at the FreeNAS/TrueNAS docs, specifically the hardware recommendations. And if you do that, you’ll notice that 15 years ago, FreeNAS required 8 GB RAM, and there was much weeping and gnashing of teeth at this onerous requirement. Up through 0.7, FreeNAS was something you might safely install on whatever castoff old hardware you might have laying around–though it still required a separate boot device (and ZFS was optional in those versions). Starting with 8.0, it became a completely different product whose hardware requirements grew substantially.

Mirror is instant, deleted/corrupted file HDD A is deleted/corrupted file HDD B. Raid is not a backup. Rsync between whilst cumbersome:
1- Buys you time in case you screwed up or corruption occurred.
2- Did Rsync fail. I get it to create a file of all files backed up. I never check it but it does, I’m sure it can also email probably.
3- Different size HDDs can be used without sacrificing space. I would love for TN to allow the use of a fuse file system ontop of ZFS such as MergerFS so you can use any size HDD without…Never going to happen so no point explaining.

That’s a load…Even so, if that’s the case then that is what makes it ideal for low end machines due to it’s robust code or is there some super duper magical dark magic code in high end machines. No, they are just built more robust. I use WD reds as they will last longer as I also tried the Seagate version which were absolute rubbish. Playing media files would jerk, same file copied to WD red perfect, back to Seagate jerking. Anything can happen on low end or high end equipment.

You Die Hards are not going to like this but let me tell. My TN Machines, two over the last 15 years have been freebies of the side of the road. First one 10 years basic use never an issue. Upgraded with other freebie for processor capabilities and liquid cooled. My Third Upgrade I already have as spare, Processor with inbuilt encryption and max of 24GB RAM.

Where I live every year there is a council pick where people put out things they don’t want. I go shopping…Whole machines, spare HDD’s, RAM, Power Supplies.
I am also salivating at what’s coming now that W10 becomes obsolete as to how many low and high end PCs will be thrown out.

So from my experience TN will run on almost anything and happily. So much so I have used Encryption from day one which is why I use TN and have had power failures whilst running and me making screw ups and TN has always recovered without issue, Over 12 years ago the batteries on the UPS died that I removed it due to how robust TN is. (Knock on wood, yep this is all going to come back and bite now that I have opened my mouth!)

Before knocking OP let them get their feet wet.

Everybody has a reason why they may or may not use TN. I use it because back then it was the only NAS with native encryption and that’s what I wanted.

To each his own.

Nobody’s knocking OP; we’re explaining, factually, what TrueNAS requires. But I’m getting more than a little tired of the recent surge in people arguing against those facts based on nothing more than their feelings.

4 Likes

Well, that’s YOUR experience, sure.

But what would you tell to those that did lose their pools? Your words will fall on their deaf ears. YOUR experience would matter none to someone who’s lost TB’s of data that usually also includes precious irreplaceable photos.

And yes, over the decades, I have amassed a sort of “greatest hits” of people that lost their pools due to janky setups both on this forum and Reddit. Most of which actually did work quite well for a year or two until some freak accidents happened. Mostly they’re unexpected power loss events that end up corrupting their meta data. One guy “accidentally knocked over” his PC; don’t ask me how that even happens; but yeah, YOUR experience is NOT the bible of truth for everyone.

The general recommendations on here is to run the most optimal setup as possible because of the tendency of a lot of beginners that would like to run TrueNAS, run it as their ONLY backup (ie. no backup) so they usually cannot afford ANY catastrophic data loss, and that risk is significantly higher with less optimal setups.

The same thing you tell a low end NAS server user as a high end trillion dollar server farm, Raid is not a backup, so backup, backup and backup for a system will fail regardless of how much money you throw at it.

Not sure what you’re misunderstanding here…
A corrupted file on member A of a mirror can be recovered and corrected from member B.
A deleted file on a mirror vdev without snaphsots cannot be recovered. But the same goes for a deleted file on a raidz.

3 Likes

Can you show me this “robust code”?

Code quality of TrueNAS is HIGHLY variable (and improving with the years, I might add). But there is code I would honestly label as “junior grade” and code I would label “Very nicely done”.

But as a whole, the codebase surely is not to be defined “robust”, in a sense of that bing a unique selling point. “about average” is what I would call it.

You honestly think the majority of people that like to run stuff cheap and cut corners would have money and will to throw on a backup strategy… or really any backups at all? I wish that were the case, then I wouldn’t keep seeing those types of posts “HALP, my pool won’t mount after fill in the blank”, but that is just not the reality.

Similar to how you say “we need backups”, we also tell them the best practices to avoid needing those backups as much as possible. They’re merely suggestions; whether one decides to follow it or not is of course, up to their discretion, and at their own risk.

1 Like

I don’t think he’s using RAIDZ.

He’s referring to this bit:

If I’m understanding this correctly, it’s just a single HDD without any siblings that’s used as a single-drive pool that gets an rsync copy every other day. Hence, the “not instant” reference. I’m guessing, he means that if there is a corrupted or lost file in the HDD A, there is a potential delay of 2 days before the corrupted/deleted file will get into HDD B.

1 Like

if the corruption is detected in time (file opened, scrub…), otherwhise rsync will just corrupt backup too

What I wrote & meant was that TrueNAS, (and ZFS), were not DESIGNED for “low end”. Either or both can RUN on the “low end”.

However, we have had people insist that USB attached disks are both good and working for them. Yet we have 2 separate forums, littered with posts of people who lost data, or entire pools because of USB attached disks. That is one aspect of “low end” that TrueNAS and ZFS are not designed to work on.

Another aspect of “low end” are single board computers. I use such as a miniature media server, WITH ZFS. It has exactly 2 SATA ports, 1 is mSATA, (a type of SSD), and the other is a 9.5mm high, 2.5" HDD drive port. Could I run TrueNAS on it? Probably, but the 4 core, 4 thread 1Ghz processor is a bit weak. Though it does have 16GBs of memory on it’s SINGLE channel memory controller. However, it works perfectly fine, serving up 1080p videos as needed.

For the last time, I am not trying to stop people from running “low end” hardware.

What I want, is for them to think about such computers, Then, IF NEEDED, (because of costs, other things, etc…), accept that it could cause problems in the future.

This convo is starting to give off winnie vibes…

make-your-truenas-dreams-come-true

I think OP likely got their answer that what they want to do is highly NOT recommended & we’ve just devolved from there… Everything is possible guys - if we got doom running on bateria, then I’m sure TrueNAS is next.

1 Like

Cheap is irrelevant and a matter of perspective, I can afford high end machines but don’t, my PC is a Lenovo Work Station from 2009 running W11 without issue and plays the games I need, Don’t fix if not broken. Cheap can mean a lot of things.

Back up strategy : My first back up strategy was a 3TB USB HDD. Copied all pool contents into it. You cannot underestimate the importance of a backup irrespective of how it’s done. Sh**t, If I have to, I’m sure I still have a floppy drive somewhere.

I understand what you are saying and true, but you can’t expect every one to have the position or finance for high end kit, It discourages a lot of would be users as a lot of would be users possibly consider home built NAS because they may not be able to afford a commercial NAS. How do most build their first NAS? with spare bits or PCs to get their feet wet.

The thing I like about this post is that everybody is right and wrong at the same time. :grinning:

Totally fair. On the other hand, I could also say that most newbies on the forums take cheap as only 1 server, no backups; and as for disaster recovery… well I’ll cross that bridge when the time comes.

Well, you obviously understand the importance of backups and not the target audience here; but I can assure you, that there are plenty of people who like to live on the edge whether it be knowingly or unknowingly out of ignorance.

And just like you say you can’t expect everyone to have the position or finance for high end kit, you also can’t expect everyone to both understand the importance of backups AND have the diligence and financial ability to fund such backup system. Given a choice between two sub-par systems (one main, one backup) and one main system (be it optimal or sub-optimal), most people would choose just a main system and cross that backup bridge when the time comes.

Haha yeap, we can drink and cheer to that for sure.