true… especially when you design your product around the ‘free’ crowdsource model…
but I can guarantee you ONE absolute truth… about EVERY user of truenas… they ARE ALL USEING ZFS … and … its ALL ABOUT THE DATA…
everything else swims around it…
its a corporate culture thing… and honestly with the multitude of open source projects I have weighed in on over the years… .this thread reminds me why I never stuck around to invest time…
Your attitude has been a bit hostile, but - ultimately - at the end of the day I agree with the gist of your argument. I’m very new to Truenas, and I’m not exactly a “power user.” But even someone as new and green as myself wants to read an actual SMART test report. Reading SMART results shouldn’t require a shell for an NAS OS, I’m sorry, but it just shouldn’t. If this were an OS for running services and whatnot maybe it might be optional, but - supposedly - that’s not what TrueNAS is meant for. First, and foremost, it’s supposed to be a NAS solution. So why can’t I check on the condition of my drives?
2 Likes
The above is insulting and inappropriate in tone. Are you used to getting your way by being a bully?
The user interface is not designed for a ZFS admin but a NAS admin. From a business standpoint, storage should be boring. If your storage is not boring then you have a serious business problem. If your storge needs an admin to be watching detailed technical statistics in order to determine if a failure is imminent then you storage technology has failed your business.
…but it nonetheless used to include considerably more information than it does now, and auto-update as well. There’s really no excuse for taking that away. Removing Netdata from the core product into an app is one thing, but removing valuable information that used to be in the core product itself is significantly user-hostile.
Sure, Netdata is pretty, but do you want TrueNAS to look like a loosely cobbled-together collection of software, or do you want it to look like a single integrated product? Saying “just use Netdata” looks very much like the former.
1 Like
not intending to get ‘hostile’ but honestly … you guys are playing pile on… but lets look at some basic retorts…
" The user interface is not designed for a ZFS admin but a NAS admin"
NAS = network attached STORAGE
Storage = in trunenas ZFS
ADMIN = to monitor manage and provision
if your NAS product doesn’t allow you to effectively administer the storage structure… in my opinion… then its missing on its core mission… when you state that is ‘enterprise’ grade…
now … if you want to be in the DROBO space … ok… I retract all my statements… feel free to design a closed box, dumbed down to mom and pop who dont know or care about storage … but then dont call it enterprise …
enterprise … without getting into real details… lets look at some basics…
todays density exceeds what… 1 petabyte in a 4u… more than 10 petabytes in a single rack… AI datacenters in particular right now are pushing density envelopes for large language models and other massive datasets…
a single 60 drive 4u storage shelf would be a real challenge to admin with this… let alone dozens of racks … without analytics, some tuning knobs (obviously some command line might be needed from time to time) … I mean just provisioning a pool… very little options are presented at the time of creation (which many are irreversible) so that an admin could provision a pool at a version and with feature sets enabled/disabled to make it compatible with another pool in the environment… without having to use command line…
anywho… I brought up a couple obersvations … and got jumped on by IX employees on how great they are because they make lots of money… we dont need your ideas…
fine… out…
One iX employee disagreed with you (in some areas reasonably, in other areas IMO not so much). If you want to play the victim about that, feel free, but that’s on you.
this thread is a fascinating trainwreck…
You compare ix to Oracle and Netapp both of which have a very strong vendor lock-in supernova that generates that revenue from large customers… they don’t offer much of anything for free and especially open because that would be counterproductive to their mission. you are comparing apples to oranges here.
you mention omni/ilumos a lot, illumos and it’s distributions always intrigued me but ilumos has such terrible hardware support that I’ve struggled to get it to run on much of anything that isn’t a VM or ancient. I have zero idea about Oracle Solaris but suspect it would be similar in this regard.
you mention oracle ZFS being “gold” but last I checked they haven’t even been forthcoming with information on potential bugs they share in-common with OpenZFS, ie the zero fill corruption bug.
as for the TrueNAS UI? I can start to agree, it needs some work, now that they have done the large changes of moving to docker and incus hopefully they can continue to refine the product as a whole for a long time now. I do miss things like ARC hit ratio, and of course I can’t even count how many times the units have been messed up here and there… but the functionality of the NAS is unaffected so I don’t really mind. I can use the CLI if I want more.
and I agree, ksmbd should probably be an option, and not default especially since kernel mode SMB modules (in any OS) have been a notorious source of vulnerabilities.
I think these conversations are valuable so long as everyone stays civil. I personally like to hear people’s views be it positive or negative but like I say so long as things stay polite and respectful. Some good points have been raised here.
1 Like
solaris is the gold standard of ZFS from both a reliability standpoint as well as what enterprise is looking for. but for many years now its been completely proprietary and the pools from their ‘version’ of ZFS are not compatible with openzfs and vis versa… for the most part… since it hard forked.
nexenta, ominos, openindiana flavors of what use to be open solaris are still considered by many to me more stable ZFS than most… again mostly due to the devs slowing things down for testing as they are all using openzfs pipeline… but just dont rush to include all the latest and greatest right away in most cases.
biggest difference again in broad strokes. is ZFS was baked into solaris at the kernel level and has tight integration for what is was really ment for… oracle database… since it runs and shares services in kernel space it has performance and integration levels that are hard to duplicate in openzfs where in most versions/distros things mostly run in user space
since openzfs moved to linux as the dev branch master so to speak… I think that linux kernel based openzfs is where it will be at going forward for the openzfs community… and canonical and others are starting to put resources behind openzfs and other tech to ensure linux supremicy in the datacenter…
my 2 cents anyway
Look. When they released 22, they made it so that when a drive breaks, the UI is incapable of telling you which one it is. On a NAS OS. The networking UI straight up didn’t work. To this day, TrueNAS does not have a simple, human-friendly overview of hardware health in your system. Again - in a NAS OS. A Windows desktop with properly set up Storage Spaces will do a better job at that. I paid $0.00 for my TrueNAS license, so who am I to complain, but when I see the state of NAS operating systems today, and the absolutely terrible decision making in their development, I don’t blame Linus for backing development of something else. It’s really tragic.
For me, TrueNAS has (had) one rather unique thing that is hard to do without their UI and it won me over years ago: I can set up VM’s and have virtual drives of those VMs protected by the same encryption as all the rest datasets. Unless I log in and manually input encryption passphrase to unlock those datasets, nothing runs and nobody will ever recover the data. I can replicate these datasets, encrypted, to different machine if needed and continue using them there provided I know the encryption passphrase. This was an absolute blessing in the field where I work. So naturally, they completely removed this in 25. I still have the latest build of Core and I keep several copies of that ISO in my vault, because to this day, it’s the most reliable, most feature complete and most secure NAS OS there is.
Why are you singling out a three-year-old release? TrueNAS has never been able to do this without iX hardware. Neither has FreeNAS. So for you to suggest that this is something that changed three years ago is frankly bizarre. And I’d argue that it’s impossible (not just difficult, impossible) to do so on arbitrary hardware–and that’s coming from someone who’s put some work into making it happen on some hardware.[1],[2] But I’d love to see code proving me wrong. I’ve suggested a method that would likely work with a bit of user input, but I don’t believe it’s possible to do it automatically.[3]
And I’d argue similarly regarding a “simple-human-friendly overview of hardware health in your system.”
3 Likes
While I agree that everything that was said could have been said with a bit more levity (wink), and I didn’t see anyone jump on anyone else, the point remains true, IMO:
TrueNAS tries to be many things, and not every design and strategy decision has been a plus, as far as I can see today.
Would I use Solaris again with some Oracle OCI instead? No.
Would I switch to poxmox/CEPH for Glusters ( 
)? Yes.
Would I keep our VMs on FreeBSD jails/bhyve? Yes.
Would I rely on FreeBSD storage hosts with ZFS and some Zabbix monitoring? Any time.
Why am I still here? Well … I don’t want to leave an old CORE comrade to die alone.
1 Like
I agree with the civil Johnny “Pull My Finger” Fartpants. 
If anyone’s wondering, here he is playing a DnD game.
3 Likes
Why are you singling out a three-year-old release?
Because I use it in production. Every release since then has been a downgrade in my opinion. Same with 25. It’s marked LTS. 3 years is not that long for “Long Term Stable.”
No version of (the community release of) TrueNAS has ever been “long term stable,” and the oldest release iX recommends for any purpose is 24.10.[1] But again, it’s your call to use a release that’s long since out of support, but the issue you cite isn’t unique to, or new to, “22” (whichever release you mean by that).
Agreed, a tight integration with the kernel and first class citizenship alongside the likes of ext, xfs and btrfs would be amazing. Not likely to happen in the near term though due to years-long licensing issues: https://www.zdnet.com/article/linus-torvalds-avoid-oracles-zfs-kernel-code-on-linux-until-litigious-larry-signs-off/
1 Like
The benefit of being able to develop ZFS independently of the OS is significant. Its a much more dynamic environment and community.
4 Likes
In what way significant and exemplified how? Afaik it’s only supported on Linux (obviously), Illumos, Windows (kind of), MacOS (kind of) and FreeBSD. As for the latter, well your own product is a good example of where the world is transitioning…
I think the history of OpenZFS demonstrates this.