TrueNAS Scale 22.02.3 Review: What Works Well and What Needs Work

[@DiStickStoffMono0xid]

Hey Tom, thank you for that in depth video! :) I've switched to SCALE a couple of months ago and i'm actually quite happy. For my use case, which is Storage, Virtualization, etc all in one box SCALE ticks most boxes. I do agree that some things are quite unfinished, but as you also mentioned, they have to catch up with years and years of optimization and bugfixing compared to CORE. Syncthing is running perfectly, yes, but the one thing that the "standard" image from IX Systems can't do is change the user group, just the user ID. (At least as far as i can remember). So I had to use the truecharts image which is also working flawlessly. Other than that i have to instances of photoprism running (via Launch Docker Container), which, after fiddling with it for some time, also works perfectly. If you do any further testing, what might be interesting for many people considering rising energy costs is power consumption compared to CORE. i did have a couple of tunables set when using core to allow the CPU to go to C3 which seems not to be possible with SCALE, or at least i'm not finding the right commands to do some optimization. It also seems like SCALE is not using the "default" power management tool but a separate one which disables all the energy saving options. Maybe you can investigate and gather some tips around that topic :) thank you! :)

[@dfgdfg_]

Thank you for the timestamps 🌟

[@robbymoeyaert7482]

Active Directory works on Scale just as well as it does on Core, including the existing issues and bugs. I will say, it's much clearer on Scale if something goes wrong with your AD integration than on Core. It's much easier to see if something is not right. And if you're wondering about the bugs, it's mainly about unjoining and re-joining TrueNAS into the same or a different AD. I've found that trying to do this is broken and usually requires a reinstall of TrueNAS to be able to join again. When I have the time to replicate the issue in a VM I will look into filing a bug report for it. The "advanced join" works fine as well. In my current setup I'm using a custom hostname, OU and dedicated AD user for my TrueNAS scale, and it works fine. The issue with the UNIX style ACLs I have encountered as well, including on importing pools with Windows style ACLs. Thing is, those ACLs are still there, you just need to switch the GUI to show Windows style ACLs after import and they pop right up again in the GUI.

[@nixxblikka]

Thank you for this video, even though I wish it would have come earlier, since for many bugs I thought it was me 😅 looking forward to the announced videos

[@_mew]

I'm very interested in you checking out the Bluefin beta! I'm very interested to see how scale advances in the future as it's what I'm running on my 140tb home server

[@clear-carbon]

thanks for confirming that these are things scale can't quite do yet, I was starting to think I was missing something!

[@UtilizatoruI]

So... I'm glad to see this pointed out. I'm looking at SCALE as a NAS option, comming from Ubuntu+Docker+extra tools mix. I was a FreeNAS user long ago but I don't really fancy BSD. That said, I have to admit that the direction SCALE took is not what I expected. My understanding is that using standard tools like rancher, helm charts, docker compose, etc to manage your kubernetes pods and docker containers is not desired by TN. For me this is almost a deal breaker since I like managing this through the above. Being forced to manually click and click and click for a mere application is annoying. It doesn't mean it doesn't make sense. It does, for those that are just getting familiar with the technology. I hope there will be more flexibility in the future since I would like a fancy NAS management system. In the end, I'm not saying they should change anything, I'm just saying that TN might no be for me.

[@sagarsriva]

Thank you. Love your videos

[@JzJad]

Have been using scale for the kubernetes management along side serving storage and asside from a memory leak issue so far so good considering I don't touch it often

[@乃-c9e]

Wanted to mention that I also checked out Wendell's video and forum on portainer. I tried it myself at the beginning of the year, before L1's posts, and hated the fact that containers run in a VM not the host. Also hate ix-systems saying BYOH but demanding using it their way. So I am using Rancher to manage k3s on TN Scale, certainly not supported but I will not accept someone else forcing their setup on my systems. Anyways, using Rancher doesn't feel too different from using Scale's UI or just straight docker in terms of manual configuration, but I make most of the decisions myself, from storage to network to images. I am very happy with my setup of containers on the Scale host, and the KVM side has been sufficient enough after 22.02, just have a couple of issues with rebooting VMs and still no USB passthrough. Additionally, I've been seeing lots of complaints about the networking and bridges, I never experienced that on Scale because I already went through all that on Core. As just a plain storage server, Core is rock solid, but for literally anything else it falls short and gives me headaches. Scale on the other hand, since v22.02, has been delivering for me. I seem to push Scale just a little more each month and it has been a fun learning experience for many things, and I'm actually using the services on it on a regular daily basis. P.S. Not hating on Core, I have two systems for my business and XCP-ng for services

[@Silent1Majority]

Looks like iX Systems got your memo and released 22.02.4 lol. Glad they did because my app disappeared and is now back. Also appears some app UI bugs fixed as well. Looking forward to your future videos for SCALE.

[@danielfisher1515]

Thanks for your informative and honest assessment of SCALE! While I find the apps work better than CORE plugins, there are still rough edges to get smoothed out.

[@potatomaaan1757]

I've switched to scale about a month ago because the realtek BSD drivers in core were corrupting my iscsi shares (thank God for ZFS snapshots) and it's been mostly good. The only bugs I've encountered are with SMB shares and ACLs, I've had some issues with those but I was mostly able to resolve all those. Really interesting video nonetheless!

[@Theborg72]

I'm running core today and have tried since I'm basically a linux user and not bsd. But I can't get it with user rights and share in scale.. looking forward to them getting easier user/group management. Thanks for the nice clip

[@deadlymarsupial1236]

One should always follow the 3 stages 1. Development 2. Test 3. Production Highly recommend import any new releases into a test environment that mirrors production.

[@d00dEEE]

So, what's wrong with having 13,220 snapshots? 😕 (That's what my system is currently showing -- and warning me about.) Virtually all of them are "prime/ix-applications/docker/006dblahblah..." so you have no idea where they came from or what they're related to, as the container id changes constantly making them completely useless.

[@willis936]

Snapshots are only as useful as they can be handled. I just wholesale delete all snapshots when I notice there is an absurd number of snapshots.

[@yusky03]

After 10,000 snapshots certain functions that you clearly have not run into will become agonizingly slow due to how snapshots are handled.

[@abdullahX001]

15:00 this is not a TrueNAS issue, this is a Docker image issue. If you have a look at the Docker page/documentation for any app, you will probably find the relevant volumes which you can mount for other apps.

[@questionablecommands9423]

I had a lot of weird hardware issues on TrueNas Core (and previously FreeNAS) the worst of which was that there were 4 disks where I would have to pull 3 of them (any 3, didn't matter which 3) to get the OS to come up. That all went away with a move to Scale.

[@mattkeith530]

.4 just came out too 🎉

[@LAWRENCESYSTEMS]

yup, got a video on that one as well

[@ReQuiem_2099]

Any thoughts on CPU arch support with weird BIG:little Intel 12th/13th gen? Excited to see Zen4 performance? Apparently even at 65W Eco mode, the 7950X is beating out the 12900K. I can't wait to see what kind of powerful, whisper quiet servers people build with them.

[@SolarianStrike]

In terms of performance per watt the 7950X beats 12900K even without eco mode.

[@ReQuiem_2099]

@@SolarianStrike Well obviously. It wouldn't beat it outright while limited to 65W if it didn't absolutely destroy it at less constrained power limits, which would still put it well below Intel's power mad 12900K. Making it, by definition, much more power efficient per performance. It's worrying that industry giants like Intel, Nvidia, AMD are just yeeting power limits into the sun, just to make "bigger bar = better" for the normies, but at least AMD is making efforts to tune for efficiency to fit their parts into reasonable power envelopes. Even if we have to wait for the "mainstream" Ryzen parts to trickle out later, THAT'S what will ultimately end up in the vast majority of big box integrator PCs for enterprise deployment. Not thousands of hopped up 250W Intel CPUs.

[@ryetzu7707]

I had a similar issue regarding the installation of the Collabora Online - Build-in CODE Server on the Docker version of Nextcloud on my Synology NAS. Whenever I installed the CODE Server onto a fresh Nextcloud instance, I was not able to reach the Nextcloud web interface anymore. I got it working again (without reinstalling), by manually deleting the CODE Server application from the filesystem of the Docker container (the folder of the CODE Server application is called "richdocuments", I deleted this folder). Afterwards I was able to reach the web interface of Nextcloud again.

[@STS]

I run my Homelab setup with TrueNAS Core U12.8. I share out x2 SSD based arrays with iSCSI to my ESX hosts. I've thought about moving to scale since this is a Homelab but I require uptime, my Homelab has developed almost a critical services side and an experimental side. Do you think TrueNAS Scale is ready for stable and performant iSCSI usage?

[@LAWRENCESYSTEMS]

The storage part seems stable but TrueNAS Core still edges it a bit for performance.

[@STS]

@@LAWRENCESYSTEMS Good to know, thanks!

I did go direct to scale and for now works fine, for nextcloud I didtn test it yet

[@yusky03]

Forgot to mention there is no way to manually bulk update or automatically update by schedule containers. So if you have 30+ containers prepare to spend 30+ min every week clicking through a slow WebUI. There is TrueTool but it updates in parallel and pulling 30+ images at the same time overloads the remote server and you get errors. ¯\_(ツ)_/¯

[@lucasgautier8373]

A risk with Scale apps is the recent deprecation of the k8s-at-home charts and repos (see issue # 1761 on their charts repo). Many of the helm charts used by iX are from the k8s-at-home project. Curious how the support for these helm charts will be handled going forward. A decent use case for Scale is strictly ZFS + KVM which is wonderful for simple home usage: a few VMs and ZFS storage for SMB/NFS shares. Unfortunately the KVM implementation in Scale does not support ballooning memory devices which leads to less than efficient memory usage across VMs. VMs in Scale lock the assigned memory from the host. Furthermore, ZFS on Linux limits max arc size to 50% of the system memory due to fragmentation in the Linux SLUB memory allocator. In my opinion, running apps that are not native to kubernetes, on kubernetes, is a bad practice. Many of these apps could be implemented in a VM with ansible and call it good enough. More Plex or Emby pods cannot be spun up for more capacity; these apps and many others being offered are not kubernetes native.

[@catsmao5313]

Valid point but personally I don't use Scale for the KVM part of it, I'm here for the gluster scale-out and better hardware support. I totally understand that a home user might not be able to afford more than 1 system ( been there done that), but it is much more effective to use a OS thats made to be a hypervisor eg proxmox instead of using a NAS OS and saying the KVM experience sucks.

[@kjeldschouten-lebbing6260]

This is inherently untrue. The Official Apps are not build based on k8s-at-home and TrueCharts is already are about 3-4x the size of what k8s-at-home used to be. All TrueCharts Apps allow for multiple pods to be deployed. If it works is another discussion, as plex, for example, does not support HA or LoadBalanced deployments featuring multiple pods afaik.

[@AdamEverythingRC]

question I am running truenas core on a asus sabertooth 990FX R2.0 amd 8 core 4ghz / 16gb mem and thinking about updating to a b550 and amd 5600 or g model. would i gain anything in speed. or should i put more money into the 990-board setup? more mem ? ect. I am using a m.2 card with a 128gb for a log cache and 5 drives in Z1. love your content!

[@neccros007]

A new end user here... Roughly how long do you think it will take iX Systems to fix all these? Not in a hurry, just curious how long issues like this take to get fixed... I might use Core but like what Scale offers with apps and VMs, etc.

[@kjeldschouten-lebbing6260]

Apps GUI is being reworked as-we-speak. The snapshot issue, which that ticket is not a very good representation of btw, is likely going to be 2024 or 2025, because the problem lays with docker not iX

[@dfgdfg_]

22.12 (December release) fixes most things mentioned

[@YeOldeTraveller]

Scale might be the future, but Core is the present that you want your Production to run on. Right now, I need the storage to be stable, and I am using Core for that. I have a separate set of systems running Kubernetes to handle my virtualization activities. I may eventually be interested in a hyperconverged solution, but I will let things develop some more. I may also eventually use the scaling feature when it is more mature as that does provide a sustainable path to building out a larger array and a better way to manage the hardware life cycle.

[@CharliePearce]

...and wouldn't you know: there's a 22.02.4 update showing in my truenas now! 🙄 No info I can find on it yet though.

[@stevenpooley393]

i currently run a esxi host for my vms, should i run truenas as a vm or replace esxi?

[@LAWRENCESYSTEMS]

TrueNAS VM is not as full featured as other platforms such as ESXI

[@210Artemka]

NextCloud is a real pain in the butt. It refuses to login for some reason (the problem with NC itself, not TrueNAS). The only way to make it work was re-install and I lost all the data I have already saved there...

[@Schlumpfpirat]

Honestly I gave up on TrueNAS and ZFS. I really wanted to love it - especially after seeing how enthusiastic you were about it. Sank in 20.000 EUR to get a proper system with 128GB RAM, 16 bay x 18TB SAS drives, proper HBAs, Supermicro Chassis + MoBo etc. But the matter of the fact is that performance just goes south after even 50% of pool utilization even when working in ideal conditions and big block sizes. Additionally iSCSI performance is simply just utter trash and seemingly unfixable due to it being a SAN protocol. It all wouldn't be so bad if deduplication would not be *highly discouraged* within the community and iXsystems too. Going with MinIO and Erasure Coding now. Veeam is making sure to let everyone know with v12 that object storage is the future, and contrary to popular belief, MinIO on TrueNAS is not a supported configuration. Aside from that I also see quite a lot of benefits with Erasure Coding in general and for scaling.

[@blackrockcity]

I never even got iSCSI to work in TrueNAS core and some people attribute that to not having enough ram. I think I gave Truenas core 20gb? I switched to NFS and it’s worked flawlessly for my needs so far. But I’m just a hobbyist compared to you. I think I spent about $500 on my system at most.

[@steveo6023]

I tried to replace Proxmox with TrueNAS Scale and it was a horrible experience. Bugs everywhere.

[@randleqgod]

Truenas Core with Ubuntu and Proxmox is a rock solid set up for me. I don’t trust scale yet 😂

[@YetAnotherToken]

Tell me more. TrueNas standalone and Proxmox in separate Host(s) running Ubuntu in top?

[@aah134-K]

I used it for 2 weeks, seemed better than expected, weird thing, my cpu show average temperature very low value, doesnt look right. also the network speed kinda not right.

[@udirt]

i know so many people let down by trying to do gluster in prod scenarios, i just feel sorry for them. i also met like one guy who had it under good control but not gonna bet anything on it.

[@kjeldschouten-lebbing6260]

Gluster on ZFS, is a different beast than standard gluster though... Performance is totally different,

[@kevinhilton8683]

Only thing that irks me was scale was in beta for so long then it was released as a final version. It's still clearly in beta. Definitely pushed out the door by the marketing people

[@essdee800]

Thanks

[@blender_wiki]

Intensively tested in a cluster setup as is supposed to become and is absolutely not ready for productions. More over is really necessary that they add an easy options to use pods instead than docker for licence and security reasons. Look like that will not be ready before 3Q 2023.

[@bryce.ferenczi]

Lawernce Systems out here doing a 22.02.3 review a day before 22.02.4 release 🙈🙈🙈

[@Felix-ve9hs]

IMHO if you dont have unsupported Hardware or if you dont need Docker/KVM/[Insert Linux only feature] you should just stick with TrueNAS CORE (for now).

[@zeron851]

My biggest problem with TrueNas as a whole(scale made this better slightly) was always how ridiculously convoluted many things needed to be for the end user. Even something as simple as creating a pool, then a share, then assigning proper permissions to the users who use that share just felt obtuse for the sake of being obtuse. I feel like having a separate toggle mode that obfuscates a lot of the complexity could help quite a bit until you get comfortable enough to *want* to introduce the complexity into your workflow. Unraid absolutely murders truenas in this regard in my opinion.

[@albertserralle6648]

You're talking as if Jails on Truenas CORE were perfect or anything... But no, they're not, you need to take into account all that same complexity you described: data vs config mounts, filesystem user, acls, etc.

[@gregoryp203]

with gluster you wont need syncthing. also the idea of running apps on a NAS seems wrong . TrueNAS core had a crappy implimentaion of virtualization so as expected True NAS scale has a crappy implimentaion of container orchestration

[@d00dEEE]

Does gluster run on iOS, Android, MacOS and Windows now? I thought it was only on Linux, but I haven't looked at it for years.

[@kjeldschouten-lebbing6260]

@@d00dEEE nah he just doesn't understand the difference in usecase between syncthing (end-user-sync) and gluster (distributed filesystem)

[@pabss3193]

It is NOT stable lol, I will say it for you, went back running to CORE

[@blackrockcity]

Thanks for the field report

[@pabss3193]

@@blackrockcity anytime

[@TechySpeaking]

First