Yet more new SCALE users needing help

[Whattteva]

I'd imagine that most CORE guides would translate almost verbatim to SCALE as long as it isn't jabout ails and bhyve VM's.

 

[winnielinnie]

I'd imagine that most CORE guides would translate almost verbatim to SCALE as long as it isn't jabout ails and bhyve VM's.

They've diverged, and Core is looking to stagnate to be left behind in my (and others') opinion.

Differences in the underlying operating system.

Differences in the ARC / memory management.

Differences in the layout and location of GUI elements.

Differences in how to access certain menus and pages (and even buttons).

Differences in user/account management (such as root vs non-root admins).

One type of ACL (permissions) in Core, yet two types available in SCALE.

Plus, of course, the paradigm shift of jails vs apps, iocage vs K3s.

 

[danb35]

One type of ACL (permissions) in Core, yet two types available in SCALE.

No, CORE has ACLs too--I don't think there's any difference there.

 

[winnielinnie]

No, CORE has ACLs too--I don't think there's any difference there.

SCALE let's you choose between POSIX and NFS4 ACLs. Core has no such choice if you want to use ACLs.

/scale/scaletutorials/datasets/permissionsscale/

When the ACL type changes from POSIX to NFSv4, internal ZFS ACLs do not migrate by default, and access ACLs encoded in posix1e extended attributes convert to native ZFS ACLs.

When the ACL type changes from NFSv4 to POSIX, native ZFS ACLs do not convert to posix1e extended attributes, but ZFS will use the native ACL for access checks.

To prevent unexpected permissions behavior, you must manually set new dataset ACLs recursively after changing the ACL type.

/scale/scaletutorials/datasets/datasetsscale/

There are two Add Dataset or Edit Dataset screen ACL settings in the Advanced Options settings that you need to configure to use ACLs, ACL Type and ACL Mode.

You must select NFSv4 in ACL Type before you can change the ACL Mode setting. The system changes the ACL Mode setting if you select POSIX in ACL Type.

Leave the ACL Type Inherit checkbox selected to preserve the ACL type from the parent dataset. For SCALE, which is based on Linux, use either NFSv4 or POSIX. Warning dialogs display after selecting either setting. NFSv4 is richer than POSIX and is used to losslessly migrate Windows-style ACLs across Active Directory domains (or stand-alone servers). POSIX ACLs are a Linux-specific ZFS feature, used when an organization data backup target does not support native NFSv4 ACLs. Since the Linux platform used POSIX for a long time, many backup products that access the server outside the SMB protocol cannot understand or preserve native NFSv4 ACLs.

That's fore SCALE.



However for Core, it's either "basic Unix permissions" or "POSIX ACLs". No means to use NFS4 ACL type.

 

[Whattteva]

They've diverged, and Core is looking to stagnate to be left behind in my (and others') opinion.

I'd agree with the stagnation part, but that's mostly on the apps side. As far as the primary NAS/ZFS functionality is concerned, I'd imagine it's still mostly on par.

Differences in the underlying operating system.

Yeah, but if you're doing things on the UI (which is what's recommended anyway), this difference doesn't really matter for most users. Moreover, most commands that pertain to running a NAS (eg. zpool, zfs, smartctl, ls, cp, mv, rsync, etc...) are basically identical.

Differences in the ARC / memory management.

No disagreement here.

Differences in the layout and location of GUI elements.

Some differences in menu placement and layout, but most of them have the same names, so you can still kinda' use the same flow as long as you're aware of the slight UI layout differences. Most notable is how SCALE stashes a lot of the top level menu under "System Settings"

Differences in how to access certain menus and pages (and even buttons).

Yep, just like I mentioned above. This is basically the primary item that would go on that SCALE guide.

Differences in user/account management (such as root vs non-root admins).

I guess I'm not too familiar with this cause on my installation, I just ignore the silly (in my opinion) Linux tendency to disable root and continue using root just like how I would on FreeBSD.

One type of ACL (permissions) in Core, yet two types available in SCALE.

This, I'm not familiar of. All I'm aware of is 1 type of ACL (typically used for SMB shares) and regular Unix permissions. I'm interested to know what the other type of ACL is.

Plus, of course, the paradigm shift of jails vs apps, iocage vs K3s.

Agreed, this is also the primary item that would go into the SCALE guide.

Overall, as I said earlier, I still think most things NAS-related like ZFS, SMART, shares, etc. from CORE would directly translate to SCALE aside from apps, jails, KVM, bhyve, etc. Heck, even HW recommendations would probably translate well even though Linux has better HW support cause a good reliable NAS system should really not use gamer gear or whatever fancy thing you can use for Linux anyway.

 

[winnielinnie]

This, I'm not familiar of. All I'm aware of is 1 type of ACL (typically used for SMB shares) and regular Unix permissions. I'm interested to know what the other type of ACL is.

See my above post. I think we posted at around the same time.

 

[danb35]

I just ignore the silly (in my opinion) Linux tendency to disable root and continue using root just like how I would on FreeBSD.

From what I've seen from (IIRC) Kris, this option is going to go away, perhaps in 23.something.

 

[Whattteva]

See my above post. I think we posted at around the same time.

Haha, we did indeed! Thanks for bringing that into light. Learn something new everyday.

 

[Whattteva]

From what I've seen from (IIRC) Kris, this option is going to go away, perhaps in 23.something.

What's going away? The option to use root at all, or the warning?

 

[danb35]

What's going away? The option to use root at all,

What I recall seeing is that the option to use root at all is going away. Hopefully the UI is fully-functional with a non-root login by then, but I don't know that I'd bet on it.

 

[Arwen]

From my reading, (some of it between the lines), is that SCALE is starting new GUI prototyping, that will eventually be back ported to Core. So I don't think Core is going to "stagnate". It is the opposite, SCALE is fumbling along, (see recent reversed packet counts in network GUI).

Further, I would think that Core will remain stable for years to come. While each serious release of SCALE might have an "adjustment period". The next serious update is likely clustering of the apps across multiple SCALE boxes. That probably won't be butter smooth. More like chunky peanut butter.

Unless Linux developers, (and I am looking at you Linus), change kernel development into stable API, (at least major releases), then Linux will ALWAYS be less stable than Core / FreeBSD.


I see SERIOUS problems in actual production. We are required to patch servers within a specific time frame. Except the unstable kernel API breaks a mandatory tool. YES, EVEN ON RHEL. This means any patch cycle that includes a kernel update, is delayed until the external vendor of this mandatory tool both updates and thoroughly certifies the tool for the new Linux kernel.

This is so bad that by the time we have the mandatory tool available for install, a new kernel update is likely available. Repeating the cycle. So in essence we are 2 to 3 months behind patching, forever.

 

[truecharts]

Further, I would think that Core will remain stable for years to come. While each serious release of SCALE might have an "adjustment period". The next serious update is likely clustering of the apps across multiple SCALE boxes. That probably won't be butter smooth. More like chunky peanut butter.

Not very likely, next major will be the GUI rework for Apps and 1-3 majors after that clustering might become the big thing.
So basically in 2-4 years, you'll have that fancy clustering ;-)

Unless Linux developers, (and I am looking at you Linus), change kernel development into stable API, (at least major releases), then Linux will ALWAYS be less stable than Core / FreeBSD.


I see SERIOUS problems in actual production. We are required to patch servers within a specific time frame. Except the unstable kernel API breaks a mandatory tool. YES, EVEN ON RHEL. This means any patch cycle that includes a kernel update, is delayed until the external vendor of this mandatory tool both updates and thoroughly certifies the tool for the new Linux kernel.

This is so bad that by the time we have the mandatory tool available for install, a new kernel update is likely available. Repeating the cycle. So in essence we are 2 to 3 months behind patching, forever.

This is less of an issue with SCALE as it:
A. Sticks with LTS, which does not have many breaking kernel changes
B. Is an appliance OS, rather than a DIY OS.

 

[Arwen]

LTS kernels are no magic bullet. I stick with them on my home systems, and still have occasional problems. That is because certain newish features are back ported, which can cause breakage.

 

[Whattteva]

LTS kernels are no magic bullet. I stick with them on my home systems, and still have occasional problems. That is because certain newish features are back ported, which can cause breakage.

I'm not too familiar with the kernel development practices, but I'd imagine they'd at least maintain ABI stability even through back-porting.

 

[Arwen]

Yes, it is very likely that a LTS kernel's back ported code does not change the API. The problem arises because you still need to recompile every single module for the new kernel. This is because their may be presets from include files that have changed. Or macros that now do something different. You might get away without doing that, but if, (or when), something strange happens, you can't be certain it was not due to the changed kernel source code.


At home to keep things utterly clean, I create new ZFS datasets for kernel updates. And in the case of changing a kernel feature, I snapshot then clone the existing kernel dataset, updating the local revision number at the end. This does force me to rebuild all packages, like OpenZFS, that have kernel modules. On the other hand, my 3 / 4 home systems have been a lot more reliable since I started doing that more than 4 years ago.

 

[RapidScampi]

Do you think it is good to have a ton of new SCALE users, even if they have to ask for help?

To be honest, it kinda riled me reading this and I was quite ready to dive in and call the post out for snobbery. Having read the thread and also seen the very welcome Beginners Intro and similar threads contributed by others in this thread I've wound my neck in

Personally, I don't think there's anything wrong with asking for support even when there's good documentation available. People are different and although laziness is undoubtedly a factor in some cases, there's nothing wrong with asking for assistance for the most part. Being able to discuss a problem is not the same as studying technical documentation and provides an important trail for others experiencing the same issues.

Ironically, I'm actually here for some support advice as the documentation (or functionality in some cases) is not there yet with Scale. I'm not a newb and I've been using Scale since the day the Alpha hit and I love it, it would be great if the documentation was fleshed out more -- something I'd also say is true about TrueCharts apps.

I say that in a constructive sense, particularly to the TrueCharts guys because I think the absence of some fundamentals - app config layout, networking dos and don'ts, storage dos and don'ts, best practices etc. as well as (critically imo) more information about changes to the aforementioned logic in the form of - changelogs etc. Standardising each release/update with a few bare essentials perhaps at the expense of having quite so many apps in the catalogue would probably lighten the support load and keep users on track.

I don't take for granted that TrueCharts is volunteer-led - the work done is hugely appreciated by everyone in the community. I also have massive respect for iX who do one of the best jobs of running an open source-led business in the game in my eyes. Just my responsive food for thought...

 

[Arwen]

I was not complaining about the "ton" of new users, just more less pointing out they exist and need / want help.

However, some of that need / want help is due to not researching or reading up on TrueNAS in general, SCALE & ZFS in particular. This leads to misconceptions of what TrueNAS is, that could have easily been answered by a few hours of research. (And I am not saying they have to do the "hours" of research all at once.)

Yes, SCALE is NOT finished. We have had dozens of new TrueNAS select SCALE, for whatever reason, find problems. Then not understanding how or why TrueNAS is developed. And what is different between Core & SCALE. We even have people thinking Core will be deprecated whence SCALE is feature complete.

The main out-come of this discussion thread is a SCALE Beginners Intro Resource:

Welcome to TrueNAS SCALE - Beginners Intro

Since mid 2021 we have had many new users of TrueNAS wanting to try SCALE. Here are some things that are helpful for new TrueNAS SCALE users to know; SCALE is a specialized & targeted OS based on a Linux distro. However, SCALE is NOT a Linux distro, and updates are by full replacement with a...

www.truenas.com

If you find things to improve for the wording, items to add or out right things wrong, simply reply with them in the Resource discussion thread, not here.

 

[Whattteva]

However, some of that need / want help is due to not researching or reading up on TrueNAS in general, SCALE & ZFS in particular. This leads to misconceptions of what TrueNAS is, that could have easily been answered by a few hours of research. (And I am not saying they have to do the "hours" of research all at once.)

To be fair, CORE users also do this, just to a far lesser extent of SCALE. I've been noticing a significant influx of beginner users of SCALE in large part, probably because the Linux base just attracts a lot more people from that demographic. And as far as the lack of research, I think you're a little bit off. I think majority of them probably did do research... just not the right type (ie. YouTube videos vs written contents on this forum).

Yes, SCALE is NOT finished. We have had dozens of new TrueNAS select SCALE, for whatever reason, find problems. Then not understanding how or why TrueNAS is developed. And what is different between Core & SCALE. We even have people thinking Core will be deprecated whence SCALE is feature complete.

Yeap, I'd say this is a result of the same root issue. Linux base just tends to attract a lot wider variety of demographics (ie. fresh homelabbers, casual Plex users, or just people who want to salvage their old gamer gear). And Linux's much wider HW support + TrueCharts just lures all these people that would otherwise be using say.... OMV or Synology.

 

[Arwen]

@Whattteva - You should probably change the attribution for those quotes from RapidScampi to me.

But, on your responses, yes, it is probably the case for the research. They are watching YouTube, or reading some blog / forum / news article saying how easy TrueNAS is. Or that now that it's Linux, it works better.

I still remember having new users come to the forums, and based on their needs & desires, suggesting something other than FreeNAS. My additional response was something like, "There was no one NAS to rule them all".

 

[Whattteva]

@Whattteva - You should probably change the attribution for those quotes from RapidScampi to me.

Whoooops, I have no idea how the quote attributions got mixed up like that. I've fixed them.