TrueNAS core 13.1 - broken UI and SMB share not reading files

Hello everyone!

I’m brand new to the forum, and I haven’t spent time in one for a long time. I’ll try to post as much useful information as I can, having tried to do some unsuccessful troubleshooting. Please don’t cringe too much, this has been a small tinkering project of mine for the last years. Also, please fasten your seatbelts, get a coffee, or a beer, and make sure you’ve paid a visit to the toilet.

First of all, it’s a VM (eww).

The metal (PLEASE READ FOOTNOTE ON HARDWARE CHANGES):

  • Intel Core i7-4790K, 4c8t @4.00Ghz
  • Asus ROG Maximus VII Gene (Z97 Chipset)
  • 24GB DDR3 non ECC RAM
  • two onboard SATA controllers:
    • one is the Intel Z97 chipset, providing 6 ports
    • one is an ASMedia ASM10621/ASM1062 (rev02), providing two ports
  • SATA HBA with another ASMedia ASM1166 (rev02) providing 6 more ports
  • Various HDDs that I don’t think are inherent to the discussion. If I’m wrong I’ll post them ASAP
  • Runs UNRaid 6.12.13 as hypervisor
  • Has a remote share mounted, which is shared by the TrueNAS machine

The VM:

  • has two virtual cores
  • 6144MB RAM
  • SATA HBA (with the ASMedia ASM1166 (rev02)) has been bound to vfio group, therefore showing the devices at boot
  • has 6x1TB SATA HDD, no cache devices.
    • Toshiba DT01ACA100 1TB
    • Hitachi HDS721010CLA632 1TB
    • Seagate ST1000DM003 1TB
    • Seagate ST1000DM003 1TB
    • WDC WD10EZEX-60ZF5A0 1TB
    • Seagate ST1000DM003 1TB
  • TrueNAS CORE version (as per uname -a output): 13.1-RELEASE-p7 n245428-79e75956dbb

I’m experiencing two separate issues on my TrueNAS VM:

first issue is that the UI (http://truenas.local/ui/) loads incorrectly.

What works:

  • I can connect to a noVNC session from the hypervisor and get a root shell, that’s how I managed to get much of the information listed here
  • Shell is apparently as responsive as it gets
  • hostname gets resolved correctly
  • browser recognizes that I’m attempting to connect to an HTTPS website of which I don’t have a certificate, asking me if I want to go on
  • the page gets loaded (loading icon stops spinning, I have the correct favicon.ico), but the only thing I see is a blue background. Browser dev tools tell me there’s an issue (This page failed to load a stylesheet from a URL. 1 source: ui/:34)

What doesn’t work:

  • I have no GUI to connect to

What I’ve tried:

Issue number le second: my SMB share, on Windows, can enumerate folders on all levels, but feels very sluggish (I know that’s as specific as “I am in North America”), and I cannot copy nor read files from it.

What works:

  • from my Android 11 mobile phone I can use VLC and RS File Explorer to access and read the SMB share (I can play music from it) with the same TrueNAS user I have in Windows
  • I can open a file from my phone, but trying to open the same file from the computer leads to “unknown format or damaged file”. I’ve tried with MP3 files, JPG photos and PDFs

What doesn’t work:

  • I cannot open, nor copy, any file from the SMB share in Windows. Any copy to local disk leads to Windows screaming “unexpected network error” or “timeout error”

What I’ve tried:

  • not much, to be honest. Google didn’t provide me many tips on this specific problem, short of restarting the machine.
  • I actually found a thread on this forum (SMB service randomly turns off due to going OOM) that apparently was correlated with my issue. I tried to give the machine 2 more GBs of RAM, bringing it to 8192MB available, but this didn’t fix the issue (not that there’s been configuration nor firmware changes that would justify this sudden increase in need, since as far as I can remember, at least 3 of those original 6GB were marked as free)

FOOTNOTE ON HARDWARE CHANGES (Or, What I may actually have broken):

due to rising energy costs, I’ve tried a few things in order to cut my power bill.
At first, I swapped out my i7-4790K (4c8t 88W TDP) with an eBay sourced i3-4170T (2c4t 35W TDP). However, it would fail POST. I believe I got a DOA chip, so I got another one in the mail, still to be opened.
Seeing this wouldn’t work (tried detaching all devices, booting with 1 stick of RAM, swapping back the old processor, clearing CMOS etc) I reinstalled my old nuclear reactor (4790K) and disabled two of the four cores, leaving two cores and four threads active. I believe I managed to re-enable all the BIOS settings.
Having now two of the four original cores, the TrueNAS VM would not start, because resource assignment was broken. I changed the assigned cores to two available virtual cores and it fired up. Unfortunately, I cannot remember if, in this status, I was already experiencing the aforementioned issues. However, in the process of troubleshooting, I’ve restored the original 4c8t configuration, and tried to give back THE EXACT SAME THREADS to the machine. This did not fix neither of the issues, but strongly alleviated the sluggishness of folder traversing.
I also tried other 2 virtual core combinations, because I am not so certain of the original virtual core assignment anymore, but I don’t believe it’s massively important.

Who I am:

I’ve been working in IT for a few years, I’ve done some LV1 and LV2 helpdesk and I’ve dabbled with VMs and Linux, but I’m more versed in Windows. I’m not afraid at all to use a CLI, but I might not know some basic troubleshooting that Linux users may be expected to do.

What I think:

I may as well have a problem with my Windows installation, but I can’t confirm it, since I have the same issue on another Windows 10 machine. I’ve been meaning to test on an Ubuntu 2404 VM, but I still haven’t gotten a round tuit. I believe those problems may be correlated, therefore I don’t know if it’s worth it to export my configuration from the CLI (if possible), nuke the machine, start fresh, load the config and import the pool. From SMART it seems the HW is OK, it’s just another case of bad computer mana and I’m all out of ideas. I’m available to provide abundance of details, if relevant, in all my few spare moments.

Thank you for your time, for reading until the end, hope you had fun and your mind is tweeked

I’m well aware there are many, if not all, things wrong with the choices I’ve made. This is, for me, more of a light hobby than a serious dedication. I understand that, if I wish to make things right, I need to get a whole new platform that supports ECC RAM, run TrueNAS on bare metal and get NAS grade drives. My journey began with unraid, an old platform and a few spare drives, then it grew to this point, with the justification to this being “what if I could do this”, instead of “the best practice is that”

If anyone is reading, I’ve managed to test the UI working on a Ubuntu 2204 LTS VM, with the default browser (Firefox 100). Curious, I decided to install the same version onto my Windows 11 machine. Another Windows 10 VM in the same host cannot access the UI neither from Edge 133 nor from the same Firefox 100.
Therefore, I only have problems with Windows machines.
My garbage NAS has apparently gotten pretty snob on me using Windows machines :grin:
Still need to work out the culprit.

According to the browser devtools, the call hangs while trying to call https://192.168.1.11/ui/assets/iconfont/mdi/css/materialdesignicons.min.css. The GET call takes about 80’000ms consistently across all the Windows machines tested.
After the call completes, another call takes an inordinate amount of time, another 67’000ms to call this file: https://192.168.1.11/ui/scripts.a6f095e3d5bfe5af0e77.js

I’m beginning to question Microsoft Defender…