TrueNAS 26.0.0-BETA.1 is Now Available!

The TrueNAS team is pleased to release TrueNAS 26-BETA.1!

:warning: Early Release Software

Early releases are intended for testing and feedback purposes. Do not use early-release software for critical tasks.

This first public release version of TrueNAS 26 has software component updates and new features that are in the polishing phase. See 26 Major Features for an overview of what’s new in this release.

:warning: Upgrading from TrueNAS 25.10

Upgrading from TrueNAS 25.10 to 26-BETA.1 is not available in the TrueNAS UI until TrueNAS 25.10.3 is released. Users on TrueNAS 25.10 who wish to test 26-BETA.1 before that time can manually install or upgrade by downloading directly:

Special thanks to (GitHub users): Franco Castillo, AquariusStar, Rogelio Tajes Piñeiro, Aurélien Sallé, dany22m, ReiKirishima, Christos Longros, Lee Jihaeng, Aui162, Seele Volleri, Ban, Michael Rohrhirsch, PCAsusM1981, Cantabile, Fernando G. Monteiro, Joda Stößer, Marius, herbkk, saso-g1, René, Jehu Marcos Herrera Puentes, Amir Burbea, Piotr Jasiek, Eric Schultz, Kent Ross, fkwp, Gautam krishna R and Joel May for contributing to TrueNAS 26-BETA.1. Visit our guide for information on how you too can contribute.

26-BETA.1 Notable Changes

  • Adds support for LXC containers in Enterprise High Availability (HA) configurations (NAS-138309). Containers can now fail over between HA controllers. HA container failover requires a static IP configuration. See Containers for configuration details.
  • Adds GPU passthrough support for LXC containers (NAS-138569, NAS-138570, NAS-138700). Users can assign NVIDIA and other supported GPU devices to LXC containers from the container configuration screen in the UI.
  • Adds Multi-Path I/O (MPIO) support for Fibre Channel connections (NAS-137252). Fibre Channel configurations can now use multiple paths for improved redundancy and throughput. This option is available in the Fibre Channel port configuration.
  • Adds SMB3 unix extensions support for multiprotocol shares (NAS-139988). When a share uses the Multi-Protocol purpose (for example, SMB combined with NFS or local app and container access), TrueNAS now enables SMB3 unix extensions. Linux clients with SMB3 POSIX support can use filesystem primitives not normally available through standard SMB semantics. Windows clients without unix extension support continue to behave normally.
  • Adds BRT (Block Reference Table) support to the zpool prefetch command for faster pool import operations (NAS-139230). Pool imports on systems that use block cloning are now faster, as the prefetch operation includes BRT metadata.
  • Adds an option to de-register a system from TrueNAS Connect (NAS-139544). Users can now remove a system’s TrueNAS Connect registration from the TrueNAS Connect configuration screen without needing to contact support.
  • Adds support for the include: key in custom app Docker Compose configurations (NAS-137498). Custom app Compose files can now reference external Compose files that define services, allowing users who manage their own Docker Compose files outside TrueNAS to use modular configurations.
  • Updates the Pools and storage screens to reflect OpenZFS 2.4 changes, including the new separation of special and dedup vdev types (NAS-138129). Pool creation and management dialogs now correctly represent the new vdev types available in OpenZFS 2.4.
  • Improves the Storage Dashboard to show the reason a pool is degraded (NAS-138613). Previously, a degraded pool indicator offered no detail on the cause. The dashboard now provides context so users can take corrective action.
  • Updates the Samba build to version 4.23 (NAS-139190). See the Samba 4.23.0 release notes for upstream changes. Note that changes to Samba defaults do not necessarily change TrueNAS defaults. See Software Component Versions for all component version updates in this release.
  • Improves touch and mobile usability for side panels and configuration screens (NAS-139925, NAS-139786, NAS-138896). Side panels now scroll correctly in mobile browsers, canvas edge spacing is improved for touch targets, and the Save button on the Add Rsync Task screen is no longer hidden on small screens.
  • Fixes TrueNAS updates failing with errors that could leave apps non-functional or set a broken boot environment as default (NAS-139794, NAS-139545). A “pool or dataset is busy” error during updates could set an incomplete boot environment as default. A separate regression also caused apps to fail to start after updating. Both issues are resolved.
  • Fixes the System > Services screen showing as empty (NAS-139571). A regression could cause the services list to appear blank on affected systems, preventing users from starting, stopping, or configuring services from the UI.
  • Fixes an issue where datasets could not be loaded in the UI (NAS-140389). A middleware issue could prevent dataset information from loading on the Datasets screen, showing an error instead of the dataset tree.
  • Fixes available space calculations for pools with special or dedup vdevs (NAS-139820). Incorrect accounting could cause available space to display inaccurate values on pools using special allocation or dedup vdevs.
  • Fixes an issue where virtual DRAID devices appeared as physical disks in the disk inventory (NAS-140344). On pools using DRAID vdevs, virtual devices could be incorrectly counted alongside physical drives, causing inaccurate disk inventory results.
  • Fixes datasets becoming unavailable after a ZFS send replication operation (NAS-139363). A ZFS issue could cause target datasets to enter an unavailable state after a send operation completed. Datasets are now accessible immediately after replication finishes.
  • Fixes a boot delay of up to 120 seconds on systems with VLAN interfaces configured for DHCP (NAS-139038). Systems using VLAN interfaces with DHCP experienced long waits during boot due to a dhcpcd configuration issue. Boot now completes without the delay.
  • Fixes an error that prevented setting secondary IP address aliases on network interfaces (NAS-139803). A KeyError: 'alias_interface_id' error could occur when saving secondary aliases in the network interface configuration.
  • Fixes the Samba Spotlight metadata service connection so that macOS Spotlight search works correctly on SMB shares (NAS-137715). The Spotlight AF_UNIX socket connection was established as a non-privileged user, causing authentication failures. The connection now runs with the correct permissions.
  • Fixes an error that prevented editing share ACLs (NAS-139535). Users attempting to modify permissions on SMB or NFS shares through the ACL editor could receive errors and be unable to save changes.
  • Fixes NFS shares showing no available actions in the Shares screen (NAS-139490). The action buttons for NFS shares could fail to render correctly, preventing users from editing or deleting NFS shares from the UI.
  • Fixes an error that prevented updating an iSCSI auth method when Mutual CHAP was selected (NAS-139397). Users could not save changes to iSCSI authorized access entries with Mutual CHAP configured.
  • Fixes USB and PCIe device passthrough to virtual machines (NAS-139045, NAS-139356). A regression in an earlier nightly build broke the ability to pass USB and PCIe devices through to VMs. Both USB and PCIe passthrough are restored in BETA.1.
  • Fixes Rsync task setup failures related to remote path validation and host key verification (NAS-139773). Remote path validation could incorrectly reject valid paths, and host key verification could fail even after accepting the key. Both issues are resolved.
  • Fixes SNMP alerts that stopped sending notifications (NAS-140259). A regression could cause SNMP alert notifications to fail silently on affected systems. SNMP monitoring integrations relying on TrueNAS alerts now receive notifications correctly.
  • Fixes the CPU reporting chart to show both per-core and total CPU usage (NAS-135633). The Reporting screen previously only showed aggregated CPU usage. Users can now view individual core utilization alongside the total.
  • Fixes UI regressions introduced by an Angular framework upgrade, including session logouts on page refresh in Firefox and broken tooltips across multiple screens (NAS-139491, NAS-139342). Firefox users were logged out unexpectedly on page refresh, and tooltips and contextual popovers stopped working throughout the interface. Both issues are resolved.
  • Fixes the TrueNAS web UI, preventing NVIDIA driver removal when the GPU has already been uninstalled (NAS-137282). When an NVIDIA GPU was physically removed, the UI did not allow removing the associated driver package. The driver can now be removed independently of hardware presence.

See the Release Notes and changelog for more details.

Documentation : TrueNAS Documentation Hub
Download : https://www.truenas.com/download-truenas-community-edition

Thank you for using TrueNAS! As always, we appreciate your feedback!

9 Likes

I have this running on my simple needs setup .. seemingly running great so far :smiling_face_with_sunglasses:

4 Likes

Working well for me as well so far, one question however does the kernel 6.18 enable support for the Intel Battlemage GPUs? I have a arc B50 and have been trying to get Ollama Immich or really anything to use it without needing a VM(cuz ram).

Moving from Nightly build of 26 to beta appears to be an issue with the following being shown in the GUI on updates.

Current version:
26.0.0-MASTER+20260406-020145
Currently installed version 26.0.0-MASTER+20260406-020145 is newer than the newest version 26.0.0-BETA.1 provided by train TrueNAS-26-BETA.

Did try manual update with the “TrueNAS-26.0.0-BETA.1.update” file, but got following error. “[EFAULT] Unable to downgrade from 26.0.0-MASTER+20260406-020145 to 26.0.0-BETA.1”

Not sure how moving from nightly builds to beta is suppose to work. Of course this is just a test setup and yes can been blown away and re-installed. Figured should be able to move to beta train?

This is expected behavior, the nighly build continues against the 26 main branch after we branch for a release e.g. BETA.1, so may have changes which are planned for a later release.

1 Like

ok. thanks for that info. Will have to do fresh install and redo the setup as import of config in this scenario can be hit and miss sometimes.

Guessing on Developer train it’s gone to Beta.1 train now? Because on the current nightly for 26 I am on (26.0.0-MASTER+20260406-02014) it doesn’t allow nightly updates to continue, actually it’s essentially bricked now and can’t change or do anything for the updates/train etc. I guess can download nightly build and maybe do manual update, so not fully bricked. Anyways all dev/test this of course.

There’s evidence that it will work, but the goal of BETA is testing.

Can disks go to standby again and remain in this state when they are not accessed?

1 Like

As I understand it, the xe driver for Battlemage GPUs basically requires Resizable BAR (ReBAR) support in order to do anything more complex than basic console output - so check with that on your hardware platform as well.

2 Likes

I guess then my question would change to what could I do to test this. So far I have tried Ollama and Immich, Ollama seems to be working with Vulkan so far :raising_hands: but Immich still cant find the GPU still working on this one.

Edit: Here is the repo I used for that if anyone was wondering GitHub - eleiton/ollama-intel-arc: Make use of Intel Arc Series GPU to Run Ollama, StableDiffusion, Whisper and Open WebUI, for image generation, speech recognition and interaction with Large Language Models (LLM). · GitHub

Yea it does confirmed work in a VM and I enabled all the BIOS options

1 Like

Congrats for another amazing release !

1 Like

Is it the same driver used for ARC dGPUs (A750, A380) too?

From what I understand the whole line now uses the Xe drivers including Alchemist.

Upgraded my backup server to the beta today. A couple of notes:

  • Active Directory worked fine before on 25.10.2 but broke after installing the beta with: Active Directory secrets file lacks an entry for this TrueNAS server.
    • A restart did not help, neither did re-saving AD options (with the same options)
    • I had to clear the config and rejoin the server to the domain to get it back in a healthy state
  • After restarting, I sometimes see: Kerberos ticket for domain is expired. Failure to […]
    • Just saving the AD config again (or trying another restart) gets AD back into a healthy state
  • After restarting, I now sometimes see a warning: [EFAULT] [org.freedesktop.systemd1.TransactionIsDestructive] (“Transaction for winbind.service/restart is destructive (local-fs-pre.target has 'stop” job queued, but ‘start’ is included in transaction).",).
    • While I’ve encountered a lot of AD issues in the last few years, this is a completely new one
    • I’m not sure if this causes any issues, as AD is still “healthy” afterwards
  • After restarting, I (sometimes/always?) see this failed job on startup: service.control
    • ID: 13
    • Arguments:
      [
        "START",
        "idmap",
        {
          "ha_propagate": false,
          "silent": false
        }
      ]
      
    • Error: [EFAULT] Service not running after start
    • Also AD-related, but not sure if this causes any issues as well (AD is Healthy)
  • Issue since 25.10 still present: “Used Space” shows up as “Unknown” for a while after restarting

Otherwise, everything seems to be working fine, so far. I’ll likely upgrade my main system to beta 1 (or 2) soon, which is also AD-joined. If I still experience the issues then, I’ll probably create one (or multiple) JIRA ticket(s). But maybe the little information already helps if someone wants to look into AD.