Started to watch the video. It’s not encouraging, IF the idea is to bring TrueNAS to the masses. Still too many buttons / options / steps, even if common pain points like ACLS may have been somewhat dealt with. I’d be somewhat curious re: what the actual customer segment they’re looking to cover with this. Competent enough to deal with hardware, BIOS, etc but not smart enough for TrueNAS?
I summarized the same concern in the earlier announcement.
1 Like
Spot on. I’d argue a better business model may be along the lines of buy a pre built NAS at the right capacity, etc and then pay for the monthly plan as part of a maintenance package.
Our agreement with HexoS is that this will not be encouraged.
There may be HexoS users that migrate to using TrueNAS natively, but if they want to maintain the HexOS service/experience, they should get support on the HexOS forum.
So, feel free to politely push them back or ignore them.
2 Likes
Geeky enough to follow LTT on YouTube but not knowing to read documentation rather than watching videos? (And LTT videos at that…)
I view LTT as aimed towards geeks still in the closet, and geeks looking for a laugh.
Also, people itching to buy stuff because someone talked to them about it.
Everyone knows the documentation is perfect and covers every single use-case that a user encounters, it’s so weird that nobody bothers to read it end to end.
If you’re an enterprise administrator it’s completely reasonable to expect you to have a good understanding of all the docs that accompany any system you maintain. However for home users the onus really should be on the system itself guiding users to the “happy path”.
There’s absolutely nothing wrong with someone taking a complicated system and simplifying it so that more people are able to use it without that additional training and understanding. A lot of folks on these forums (Not necessarily yourself) are needlessly gatekeepy, like if you’re not an expert in Linux you don’t deserve to have a NAS built with commodity hardware.
HexOS is a brilliant idea and I’m all for it. Kudos to iX for recognising the need for a partner in making TrueNAS more accessible to all. Everyone will benefit from this.
2 Likes
@kris and I talked a bit about HexOS on the latest podcast episode:
HexOS, RAIDZ Expansion, New GPUs, and A Brief History of TrueNAS | TrueNAS Tech Talk (T3) E006
I’ll restate what I said there - it’s not for me and that’s fine. You shouldn’t have to be me in order to benefit from the awesomeness of OpenZFS. 
2 Likes
@Kushan - While I agree with you in general, HexOS needs to be able to supply diagnostic information to the user. Perhaps in a way the HexOS forums can use to assist in fixing the problem.
For TrueNAS, if something goes wrong unexpectedly, we resort to the Unix command line. Getting things like;
- zpool status
- zfs list -t all -r
- lsblk
- smartctl -a
- Others?
What I mean, is that TrueNAS does not support in depth GUI or TUI based troubleshooting. So it is expected that everyone will eventually need to use the Unix Shell for further data gathering or troubleshooting.
HexOS can / should be made to have the diagnostic information supplied by the GUI. How that is displayed, well, it is a judgement call in GUI design.
ZFS command lines were designed to be straight forward. But, since the Oracle Solaris ZFS and OpenZFS split, hundreds of changes have been made in OpenZFS. Those changes reduce the ability to easily trouble shoot some things.
I wish the zpool command was split. Perhaps a zmaint command to perform less common procedures. Like RAID-Zx expansion or upgrade.
That is exactly the same sort of thing I was saying.
What they are doing is making the basic setup and monitoring tasks simpler so you need less skills, but when something goes wrong these people with less skills will not have a clue what to do unless the UI has workflows for at least the top 10 most common technical problems.
1 Like
I’d say start with a few, like 2 or 3. Fine tune them. Figure out what problems the users are experiencing and add as appropriate.
One of the things we get here, is weird, or purposefully chosen bad decisions and expect it to work. Like USB attached storage for data pools. Or trying to fix something that is not broke, like RAID-Zx expansion pausing when a failed disk is found.
If HexOS can steer users to sane configurations, that may eliminate lots of problems. I mean the users are cattle, (we steer them :-), but no need to immediately slaughter them.
1 Like
I always likened my sysadmin days more to herding cats.
3 Likes
Meow, Meow
Anything that can be displayed via a CLI can be displayed in a GUI. Of course you are correct, this information needs to be presented to a user in such a way that they’re able to make sense of it.
For most users, I don’t think the majority of issues they’ll face are anything unusual in the PC space. A hardware component might malfunction, a hard drive will die - as long as HexOS is clear on “This hard drive is behaving badly and you need to replace it or risk losing everything”, it’ll be fine in most cases.
I imagine they can even bake in a lot of self-healing. I mean I’m not that old but I remember when we used to have to manually defrag our own hard drives but modern systems do that for you - I don’t see why HexOS should be any different in this case, with the majority of maintenance tasks.
The real trick will be anyone who wants to do some performance tuning but again I think the workloads for this kind of user are vastly different to a typical enterprise or even business case.
At least some of those commands getting a GUI counterpart in TrueNAS is long overdue.
1 Like
Yes - I agree that it is much better to prevent inexperienced users from doing something stupid (which is unfortunately relatively common) than to provide automated workflows to fix the consequences.
That said, adding UI elements to steer users away from doing something stupid is not that difficult - and it could be done in TrueNAS relatively easily without being overly intrusive. So why do we need a separate HexOS? We already have tooltips, and in many cases “are you sure” tickboxes - they just need to be more informative and written in terms that both inexperienced and knowledgeable users can understand.
And I am unclear exactly what will HexOS not provide a UI for (thus steering people away): 32 x 20TB RAIDZ1 data vDevs? Mixed RAIDZ / Mirror multi-vDevs? All (or most) types of non-data vDev? Unmirrored special allocation vDevs? I have no idea.
If I was to say “as long as TrueNAS is clear on “This hard drive is behaving badly and you need to replace it or risk losing everything”, it’ll be fine in most cases.” that would probably be equally true.
And I think it is perfectly possible for TrueNAS to provide statements that are simple, clear and helpful for new users without being overly patronising to experienced users.
I don’t have to go out and buy a paid-for license for a special Linux distribution to get automated defrag or updates. And IMO TrueNAS already provides a lot of this by easily allowing you to run Smart tests and Scrubs (which due to their performance hits might need to be scheduled for quiet times). That said, TrueNAS could create these as standard for each pool and suggest that the times and frequencies are reviewed, but doing this doesn’t require an entirely new HexOS either.
From what I have been able to gather since I have been using TrueNAS, most people’s attempts at performance tweaks such as adding L2ARC or SLOG are done without understanding the use cases where they are actually beneficial and are often of little benefit or possibly even negative benefit. It wouldn’t be difficult for TrueNAS to check that the system has 128GB of memory or more and provide an Are-You-Sure pop-up with an explanation if someone is attempting to add an L2ARC on a system with less memory than this. Or similarly an “Are-You-Sure because these are the impacts and you cannot undo this” when adding a Special Allocation vDev or a Dedup vDev or a 2nd RAIDZ data vDev or …
Given the infrequency that these sorts of one-way operations are done even in large scale Enterprise systems, such popups would not be overly intrusive.
The only other forms of performance tuning are by tweaking ZFS tuneable parameters by echoing a value to a system configuration pseudo-file. To date, I have yet to see anyone show that such tweaks give a material overall boost to performance - as far as I can tell from what people say here, the ZFS defaults seem to work pretty well, and those considering such tweaks are looking to improve the performance of some artificial and extreme benchmark workload (often without understanding what is happening under the covers and so without understanding what these tweaks do) and without properly testing the overall impact on their real workloads.
And I haven’t seen any generally applicable ZFS or TrueNAS performance tweaks on the internet (and Windows abounds with them) - aside from the outdated one from when ZFS ARC was limited to 50% of memory.
So I question whether there is actually a need for any performance tuning automation. (I genuinely don’t know the answer - if anyone has any examples I think we might all be interested.)
If HexOS programmes a way to analyse and identify such tweaks, then great, but I would argue that these will have little effect for the target audience of HexOS and would likely make more sense in a large-scale Enterprise setting.
So, I don’t see HexOS offering automated performance tweaks, and if HexOS or TrueNAS users want to go off piste to do this there is nothing stopping them.
Summary
Don’t get me wrong - the remote management functionalities of HexOS are definitely useful and firmly aimed at the smaller TrueNAS user, and this is NOT iX’s market segment - so a paid HexOS seems like a good idea here.
And there is nothing wrong with HexOS directing users towards good setup choices - but equally IMO TrueNAS should be directing users towards good setup choices and away from bad ones.
My concern is that the existing issue with non-technical TrueNAS users being unable to deal with drive failures or pool corruption on their own is unlikely to be solved by HexOS, but is likely to result in many more non-technical users finding themselves in this situation (assuming that HexOS drives higher numbers of users rather than their users only being people switching from TrueNAS to HexOS).
1 Like
Their is one useful performance tweak, the RAID-Zx expansion speedup. It’s covered here;
However, @Protopia your point about defaults is also valid. The average HexOS user can wait until the feature / fix is implemented into OpenZFS, then into TrueNAS and finally into HexOS.
If a HexOS user wants “cutting edge” features and fixes, why are they using either HexOS or TrueNAS? Neither support such out of the box.
1 Like
That’s a good summary of the strategy.
HexOS only supports a small % of TrueNAS features (e.g no AD) as well. That has a double benefit:
-
Those features and configs are well tested by Hexos
-
Users will have less configuration mistakes that need to be troubleshoot
Systems will be smaller (and approved by HexoS).
However, hardware failures will still happen…drive failures can be handled, but server failures will still be difficult. With 100,000 users, I’d estimate they will have 10-30 per day.
HexOS team has to learn to help those users. With 100,000 users they can afford it. The good thing with ZFS is the drives can be installed in a replacement unit.
3 Likes
…and it does this in the case of scrubs, at least last time I checked. The frequency’s a bit low for my taste, but they’re there.
1 Like