TrueNAS and Other Custom NAS Builds

[Leesureone]

Awesome question and I'm glad someone raised it!

Short answer: Yes

The washer and the dryer will report their status via iOS notification, which even my Apple Watch receives! It reports remaining time and allows for remotely setting washing and drying parameters.

View attachment 574886

That's is just plain awesome.

 

[CaseySJ]

While there are good reasons for having offsite storage, this isn’t one of them:

Google Drive users complain of missing files, months of data disappearing

Cloud storage service Google Drive is losing user files, and engineers warn users to avoid making changes or troubleshooting while the issue is investigated and resolved.

appleinsider.com

 

[CaseySJ]

Earlier in this thread I posted a photo of the EverSolo DMP-A6 streamer. It has an internal M.2 slot for an NVMe SSD up to 4TB (or maybe 8TB). I'm currently using a 2TB NVMe SSD with my entire music collection in (a) Apple Lossless and (b) DSD formats.

With the latest EverSolo software update, there is beta support for Jellyfin and Plex.

It was very easy to connect the EverSolo to Jellyfin running on TrueNAS Scale. EverSolo provides a rich interface for Jellyfin that includes cover art and various other views.

This just keeps getting better!

New firmware v1.2.30​

Changelog:

1. Optimized the occasional issue with WebDAV not displaying lists.
2. Optimized the display of playback details in EMBY.
3. Optimized the display of playback details in PLEX.
4. Optimized the display of playback details in music services.
5. Optimized the problem where track information couldn't be displayed when playing tracks via the file manager.
6. Optimized system language compatibility.
7. Optimized online DSD audio playback.
8. Optimized the occasional issue with multi-channel DSD not playing.
9. Optimized the issue where the playback queue order was incorrect during random playback
10. Added TIDAL MAX support (require re-logging into Tidal account)
11. Optimized online APE track playback.
12. Optimized the problem where the playback queue wasn't saved after device restart.
13. Optimized the occasional application crash when opening artists in the Android control app.
14. Added Jellyfin cloud drive functionality.
15. Optimized the occasional incorrect display of the bitrate of the currently playing track.
16. Optimized the display of large font mode in playback details.
17. Added settings for the highest sampling rate for optical output.
18. Added settings for USB DAC sampling rate switching delay.
19. Optimized the issue where certain specific sample rate tracks couldn't be played.

 

[Say Heading]

scrubbing and resync to HFS?

Inspired by this thread and a video about the installation of unRAID I decided to install trueNAS CORE for testing on available hardware (the Z68 in my signature). So far pretty nice!
As I´m generally in the same team as @zipb using DAS, JBODing directly within the tower cases of my builds, I see the upcoming problem to handle this properly once silicon macs will take over.
At the moment I use 2TB HDDs in JBOD, backing up via CCC to separate HDDs. Exchanging a corrupted drive 1:1 and immediately continue (never happened ... yet) - it is soo simple.
But lets be honest: This backup technique does in no way make sure that the backup (nor the older original) files are uncorrupted. Up to now I just ignored.

But with trueNAS I stumbled across scrubbing and resyncing and I wonder if I could also use this for backing up the data on the NAS to separate HDDs. With a lot of 2TB disks lying around I would like to stay in my old habit.
My shares would be organised to reflect the previous structure of 2TB disks and I´d like to backup these onto real ones. But preferably as HFS so the backups could be directly attached to the mac(s) in case of trouble, not having to rebuild the NAS data in the first place.

Can I get the advantages of scrubbing and resyncing in combination with HFS?

 

[etorix]

Scrub is a ZFS feature, which is only possible for ZFS pools.
You can install OpenZFS on OSX, and manage ZFS pools within OS X if you're handy with the command line.
Note that scrubbing can only repair corrupt data if there's enough redundancy in the pool—not with JBOD, and not with single drives (unless you're using 'copies=2'). Without redundancy, scrubbing can only report that "file X is corrupted".

I'm unsure what you mean by "resync".
'rsync' is a standard Unix utility—and it is available in macOS as well: Just use Terminal.
If you mean "replication", which is a super-cool feature, and much more efficient than rsync, this is also a ZFS-only feature, and only possible if both the source and the destination use ZFS.

I see the upcoming problem to handle this properly once silicon macs will take over.

The easiest way forward, with a view to switching over to AppleSilicon, is to move your storage to a NAS with 10 Gb/s (or faster) networking.

 

[Say Heading]

…sorry, I assume I´m at the very bottom of a steep learning curve. Thanks for your input!


My hope was to somehow profit from the cool srubbing functionality on the NAS even on my HFS backup disks. But obviously the only way to do so - without changing those backups to zfs - would be to copy those scrubbed shares 1:1 during each backup cycle e.g. by CCC and over the network.


Mine is only 1Gb/s and not easily upgradeable, but so far that speed is fine for the general purposes of the NAS at the moment. But a nightmare for such a 2TB full backup and we´re talking about several.


I´ll have to dig deeper, a very inspiring topic to deal with during the dark winter season.

 

[etorix]

ZFS unquestionable has a learning curve. The key is to sort out what's necessary for a basic use from the many advanced features which do not apply outside of large enterprise deployments. And it may be fair to conclude that ZFS is overkill for your home use.

My hope was to somehow profit from the cool srubbing functionality on the NAS even on my HFS backup disks. But obviously the only way to do so - without changing those backups to zfs - would be to copy those scrubbed shares 1:1 during each backup cycle e.g. by CCC and over the network.

Scrubbing relies on checksums to identify data corruption, so it absolutely requires a file system which checksums data blocks (APFS checksums metadata to ensure its consistency, but not the data itself).
Copying whole disks only works if the drive content does not change after the cloning. And, if you were to identify that a disk and its clone no longer match, how would you know which drive holds the correct data and which drive had a case of "data rot"? You'd need a third copy for that, to decide by majority—or a checksum of the known good data.

 

[CaseySJ]

Spent much of the weekend setting up three things:

• Home Assistant (Supervisor) on Orange Pi 5 Plus with Debian 11 (Bullseye) on a NVMe SSD
  • This controls all sorts of smart devices
• HomeSeer HS4 Pro on the same Orange Pi 5 Plus with Debian 11
  • I've been using HomeSeer for more than 10 years on various mini PCs (x86 Windows), but recently moved to ARM64 on Orange Pi 5 Plus
  • This controls the entire Z-Wave network of devices
  • Will migrate the Z-Wave network to Home Assistant soon, but only once I'm comfortable with it
    • Already installed Z-Wave JS UI to get acquainted
• Experimenting with different Dashboards
  • Homearr
  • Heimdall

Home Assistant:

HomeSeer HS4 Pro:

Heimdall Dashboard:

Homearr Dashboard:

I have both the new Raspberry Pi 5 and Orange Pi 5 Plus. After a fair bit of experience with both, I can say that the Orange Pi 5 Plus is the hands-down winner across the board. Yes its form factor a just a bit larger than the Pi's, but for that you get vastly more:

• Noticeably faster CPU performance (8-core 64-bit architecture, 4*Cortex-A76 + 4*Cortex-A55)
• Noticeably faster GPU performance (ARM Mali-G610)
• Two 2.5GHz Ethernet ports
• 2 HDMI 2.1 outputs, 1 HDMI input
  • Supports 120Hz refresh on my 4K mini-LED monitor
• NVMe SSD Slot (PCIe 3.0 x4)
• M.2 card for WiFi/BT module (I'm using WiFi 6 card)
• 16GB memory (new 32GB models coming soon)

 

[oli.mathieu]

Hello
I wish to build a NAS for Video Editing for 4 Computers with their own Ethernet 10Gb/s connection

Here is the list of parts I picked

Case : SilverStone Technology CS382
8 mechanical hard drives
Motherboard : ASUS PRIME B760M-A-CSM
CPU : intel i5 12500
Ram : Crucial Pro RAM 64 Go Kit (2x32Go) DDR5 5600MHz
Nvme for L2ARC : WD_BLACK SN850X 4To
NIC : PCIe 3.0 8X Quad ETHERNET RJ45 10G 5G 2.5G 1G 4 Ports - Intel X710-T4 CHIPSET
For more SATA ports
Nvme to Sata : M.2 to SATA 3.0 6Gbps 5 Ports
or
PCIe to Sata : PC Donkey PCIE3.0, 4 Ports

Is there any obvious bottle neck ?
Any other advice ?

Thanks

 

[etorix]

"L2ARC" implies ZFS. The 5-SATA-ports-on-M.2 thingie is NOT suitable for use with ZFS: Dubious controller, and likely SATA port multiplier, which is an absolute no-go!
Get a LSI SAS HBA, or rather a motherboard with 8 SATA ports. Since this looks like a professional build (at least for a SME), why not go professional with a genuine server motherboard?

The 4-port NIC looks like overkill, and if bought on AliExpress there's a non-zero chance to end up with a fake Intel card. One port and a switch is a safer option than four direct connections to four clients.
You are not going to saturate a 10G link with eight spinning drives—and certainly not going to be happy with direct editing on the NAS if that's the plan.

What's the L2ARC intended for? This is a read cache: It's not going to help with writes, and it's likely too small to hold all of the raw rushes of a project as a working set. Worse of all, 64 GB RAM is not enough to support that big a L2ARC (maximum: 5 to 10 times the RAM). As planned, a 4 TB L2ARC would negatively affect performance by evicting ARC (the "Level 1 cache").