As I understand it, the middlewared implements an API, and the web ui just consumes the API, as does the cli ui. And I guess as could hexos ui.
2 Likes
Linus isn’t the one writing the code, or developing HexOS. He invested in the company that is building it, that company being founded by what is probably the most popular consumer NAS software out there (unRAID).
Again, the team behind it have plenty of experience writing NAS software. You might not think much of unRAID, but you can’t deny that they’ve had years and years of experience building NAS software designed for home users.
Now we’re getting somewhere. What you’re describing is exactly why HexOS exists, because nobody’s done this right yet, nobody’s build a NAS software that’s both as powerful as TrueNAS but as user friendly. However, there are people in this thread saying “ZFS isn’t suitable for home users” or already blaming issues on inadequate hardware. Inadequate for who, exactly?
Home use is completely different to enterprise use , where lets face it you don’t really need ECC, you don’t really need an intel Xeon and you probably won’t notice a performance issue with only 32GB of RAM while still running some apps - sacrilege I know, but people are using TrueNAS like that today because their use case simply isn’t catered for. The options out there for a home user are surprisingly slim.
People keep saying TrueNAS isn’t designed for home use, people keep slamming amateur users because they made mistakes and don’t have a degree in filesystems but there’s a very clear gap for those that want actual good NAS software on hardware of their own choosing that’s more than just a bunch of cobbled together scripts and esoteric parity schemes.
Like it or not, ZFS is now mainstream. It has broken out of the Enterprise world, which is why you’re getting features like RAIDz expansion - a feature with somewhat limited value in the Enterprise where you’ll almost certainly configure the entire server from the start but very valuable to a home user who’ll grow over time. iX’s own contributions have made HexOS possible, there’s no doubt about it.
I’ve said it before and I’ll say it again, nice and clear for those in the back: If you think “ZFS is too complicated” or “users are going to make mistakes” then that’s exactly the point of HexOS. Will it succeed? Who knows! But I can’t wait to find out.
9 Likes
Leaving aside whether unRAID is actually that popular (which I seriously question), no, HexOS isn’t founded by them–it’s founded by two of their (former) developers. And for some reason they aren’t basing it on the software they’re most familiar with.
Really? Who? You’re the only one I’ve seen mention “home users.”
That depends on how important your data is. I think there’s a strong argument that it’s even more important for the home than in the enterprise, as the latter is more likely to have a decent backup strategy in place. But your family photos or home videos are likely irreplaceable, and these days they’re likely to be digital–so if the data’s corrupted, that’s the only copy.
Likewise, Realtek NICs are garbage–it doesn’t matter whether you’re at home or at work. It’s neither difficult nor expensive to avoid them, but you need to know what you’re looking for.
Again, who’s saying otherwise? It’s looking very much as though you’re arguing against a straw man.
You’re the only one saying this.
…and nobody’s doing this.
If you’d like to participate in the discussion that’s actually going on here, feel free. But it sounds as though you’d rather rebut positions that nobody here is taking, while attributing ill motives to those who are actually participating in this discussion.
3 Likes
I do say OpenZFS is complicated, especially if you take all the features in mind.
However, if a user friendly GUI makes sensible defaults and skips some of the more odd features, (at least in the beginning), then it may be possible to have a successful alternative GUI & NAS to TrueNAS, (Core or SCALE).
As for what I mean by sensible defaults, any user reasonably familiar with OpenZFS would understand what I mean. But others would argue points. Like De-Dup at home. Really? Maybe Fast De-Dup, but that is not going to be well tested for at least 1 year.
Or how about Special vDevs for Metadata or small files. The only rational defaults is to have the Special vDev’s redundancy match the Data vDevs. So a RAID-Z2 would mean a 3 way Mirror for the Special vDev. Yet lots of people might argue the point and say a 2 way Mirror is good enough. (Well, until they loose their entire pool because of it.)
In essence, people will do what they want. Including software developers for NAS products. That is somewhat the point of an open, competitive economy and not monopoly.
I think most of us here in the TrueNAS forums want people to have successful experiences with their NASes. And if TrueNAS does not seem to fit their needs, (maybe they want a Raspberry Pi style NAS), then we, (at least I), wish them luck.
Part of the problems I have seen here in the forums is that when an inexperienced user has problems, they tend to not supply useful information. They literally have no clue about ZFS trouble shooting. Then we have to walk them through how to get useful information. Something as simple as zpool status seems beyond them, (the inexperienced user). That said, if the HexOS GUI has a clearer display of the pool layout, GREAT!!! But with TrueNAS we resort to command line quite quickly.
3 Likes
This thread is going off the rails, so let’s all make an effort to stick to the topic at hand. I - and I assume most here - am not particularly keen on endless discussions that never have any conclusion because there is nothing new to discuss, only the same arguments from 10 years ago, and because people’s points of view are radically different.
So please drop the topics of “ECC”, “Hardware”, “Shortcomings, real and perceived, of the TrueNAS Community”, as far as they are not directly related to… whatever comes out of this thing.
3 Likes
I’m not replying [to a deleted post] because the mod said not to.
(Mod note: edited for cleanup)
6 Likes
Likewise, pausing only to note that the plural of anecdote is not data.
3 Likes
I hate to delete posts, so please make things easier for me and don’t respond to posts trying to derail the conversation.
And, of course, don’t try to derail the conversation.
5 Likes
VIM is the best text editor. Fight me.
10 Likes
So, getting back on topic…
I had another look at the screenshots and noticed a bit of a discrepancy, or possibly not.
On the right hand side it says 32 GB DDR5, yet the middle section says “AMD Ryzen 7 7700X - 16 GB RAM”.
This is one screenshot, so I’m wondering if that little “Attention Required” is something to do with the 16GB/32GB discrepancy (though I’m not sure what would cause that) or if it’s actually an oversight and this screenshot isn’t really a screenshot at all but a mockup.
Tinfoil hats on, I’m looking forward to people discussing if it’s a UI glitch, a system error, a conspiracy to fake an entire OS, Linus dropping half your RAM, etc.
1 Like
When I checked the screenshots i thought they were mockups as every usage dial had the same percent displayed.
But I figured that’s okay, so didn’t mention it.
Looking forward to seeing whatever it is
3 Likes
Ordinarily, I’d say “mockup for sure”. But the way DDR5 systems have been known to behave, I wouldn’t be surprised to see a legitimate hardware/memory training issue leading to half the memory being unusable (but visible via SPD on the SMBus).
2 Likes
Are you being a troublemaker? 
I’ll throw an idea out there. Linus has a perk that makes him good, he has a lot of money through his business (or so it seems).
Given that, he can bulldoze through “shooting” himself in the foot through the sheer power of money. He probably learns too through that.
If he is a good sponsor for a NAS OS, time will tell. I have no idea, but I’m not going to be a beta customer with all my data.
That being said, I learned orbital mechanics by overengineering my spaceships in KSP because I’m a terrible pilot (and a terrible engineer, but you can add more parts for as long as your GPU can handle them), so I understand the idea of “bruteforcing it”.
That’s my 2 cents
2 Likes
I hope this doesn’t derail the thread: Why?
What’s in it for ixSystems? Supporting free users and the features they want is clearly useful: Marketing, mindshare, a certain amount of free beta testing.
But prioritizing non-paying users over paying ones - at some point you’re killing off the business.
1 Like
I apologise if I gave the wrong impression…
Firstly, generally speaking ixSystems does recognise the needs of small users. Enterprise data centres are probably not the main users of virtualisation or apps running under TrueNAS - enterprises will have separate infrastructure for these use cases. I suppose some Enterprises might use apps and VMs in a branch environment to avoid having multiple servers in each branch, but they could do this using Proxmox if need be. So IMO apps and VM functionality is provided by iX for the home and small-business demographic - and whilst some of these may buy the smaller iX hardware and contribute some revenue, the majority of the users of these technologies are going to be non-paying users. In other words, if iX was simply maximising profits, TrueNAS would be a pure NAS without apps or VMs focussed purely on Enterprise customers (and likely NOT open source).
Indeed, if you look at the back-history of iX and of the key team members, they have spent decades supporting open-source and community i.e. non-paying users.
Secondly, as I have pointed out elsewhere, non-paying users do still contribute a lot to iXsystems success. For example, I suspect that the vast majority of beta testers are non-paying users - almost certainly those who are testing it in production (rather than a quick 1-day install / basic-test) are likely non-paying users because most Enterprise customers are NOT going to use a Beta version in production. I suspect that the mass of non-paying users also push more boundaries than Enterprise customers and therefore find and report bugs which are then fixed before Enterprise customers ever experience them.
Also as I have pointed out elsewhere, it is the paying customers who keep the lights on - so it is completely understandable that Enterprise customers’ needs need to be high priority.
However, the point I was trying to make was that iX should still make reasonable attempts to keep the non-paying users happy - and not screw them over when prioritising Enterprise functionality - because if it gets too painful, then someone will fork your code and create a competing offering, thus splitting the user base into factions and history suggests that this leads first to inconsistencies and then development efforts being duplicated and diluted and eventually to one fork dying away - and sometimes if the pain is too quick and too great, users disappear overnight.
As a side note, I’m not sure I trust LTT enough to put them inside my NAS.
1 Like
Once again, LTT are not behind HexOS. Linus Sebastian has invested in the company behind it but that’s it.
2 Likes
Admittedly I did not read all the 118 posts, so you are saying they are just investing and not providing any prompt to the development of this… Hex OS?
1 Like
Here is my reading (i.e. educated guess) about this:
-
LTT are always looking for publicity opportunities and they see one by promoting a more friendly NAS for non- or less-technical users. I have no idea just how committed they are to delivering it - especially if they have to keep increasing their investment.
-
I would assume that the Unraid engineers they are hiring also believe that Unraid’s UI is not good enough either.
-
I am unclear just how they are going to get revenue to pay back the investment - either by selling licenses (for their added value code) like Unraid or by selling hardware on the back of the software (which will cut the global pie into smaller slices - but might be a partnership opportunity for LTT to make some money).
-
They would I think have preferred to start by forking Unraid, but it isn’t open source. So rather than start from scratch, they have decided to fork TrueNAS which is open source (and LGPL rather than GPL, so they can license their added-value whilst still distributing TrueNAS).
-
I am not confident that it is going to be an easy ride technically to integrate with TrueNAS’ middleware.
-
There is a fundamental issue with making the UI easy for non-technical users. It is (relatively) easy to create an easy to use simplified UI for non-technical people to do normal operations i.e. create a pool. (Really good UX is never easy.) But if you are going to make it usable by non-technical users, then you have to make that the case when things go wrong - not only single things but also combinations of things. Not only is the number of single error workflows one or two orders of magnitude more numerous than the normal workflows, handling combinations of errors multiplies these up still further. So you have to enumerate all the possible errors that could occur, and the combinations of these, and then you have to ensure that your diagnosis algorithms don’t create false positives or false negatives, and you have to create robust recovery code that will definitely fix the error and not make things worse, and handle any failures of this recovery code etc.
In other words, making a NAS require less technical skill and making it fool-proof and able to handle errors is a massive undertaking.
-
Although the ex-Unraid staff will have excellent general NAS, and Unraid experience, Unraid only started supporting ZFS this time last year, so their ZFS administration experience may be limited (which is an issue for TN which is ZFS only). And there will be a significant learning curve to understand how to add their own UI on top of the TrueNAS middleware and also to enhance the middleware to cover use cases it doesn’t yet handle.
In summary, there is a HUGE difference between publicising a launch and showing a few (admittedly nice) wire-frame UI graphics for a basic use case, and actually delivering a comprehensive and robust UI for non-technical people on top of someone else’s middleware (middleware which you didn’t develop yourself, which will be a moving target, and which doesn’t support all the use cases you need to cover etc.).
I think we just need to wait and see what actually gets delivered.
1 Like