The TrueNAS team is pleased to release TrueNAS 25.10-BETA.1!
This first public release version of TrueNAS 25.10 (Goldeye) has software component updates and new features that are in the polishing phase.
Early Release Software
Early releases of a major version are intended for testing and feedback purposes only. Do not use early-release software for critical tasks.
25.10-BETA.1 Notable changes
Introduces a redesigned Updates screen that allows users to select an update profile reflecting their risk-tolerance level (NAS-133600). TrueNAS only notifies users of updates recommended for their selected profile level.
Introduces changes to Certificates, Certificate Authorities (CA), and Certificate Signing Requests (CSR) (NAS-135168):
Removes Certificate Authorities (CA) screens and support for CAs, which means you cannot sign CSRs or create self-signed certificates.
Removes Add from Certificates. Users can import a certificate created by an external certificate authority.
Allows adding a certificate using the new Create ACME Certificate screen found under Certificate Signing Requests and an existing DNS authenticator added through the ACME DNS Authenticators screen.
See Preparing to Upgrade for more information.
Introduces a refreshed Users (WIP) screen that provides improved, at-a-glance access to account information and simplified user creation (NAS-134198).
Introduces changes to the Datasets and Storage Dashboard screens (NAS-135362, NAS-135364).
Renames dataset and pool widgets, and changes links to other screens.
Removes Scrub Tasks configuration and scheduling from the Data Protection Tasks screen, but makes it available on the Storage Health widget located on the Storage Dashboard (NAS-135555).
Fixes the NVIDIA GPU related error “RenderError: Expected [uuid] to be set for GPU in slot” (NAS-134152).
Adds support for for directory services authentication via FTP (NAS-135200).
Adds the Enable Secure Boot option to virtual machine configuration screens (NAS-136466).
Allows TrueNAS to automatically migrate existing applications when selecting a new applications pool (NAS-132188).
Adds TrueNAS Apps service support for configuring external container registry mirrors as alternative sources for Docker images (NAS-136553).
Introduces various UI improvements and optimizations simplify core user experiences (NAS-135159).
Includes critical ZFS stability fixes and performance improvements, including fixed corruptions for plaintext replication of encrypted snapshots, enhanced memory pressure handling, faster pool export operations, improved I/O scaling capabilities, zfs rewrite and Direct I/O support (NAS-135902).
Simplifies and improves robustness of gateway and name server settings when changing from DHCP to static aliases (NAS-136360 and NAS-136360).
Moves Network screen under the System main menu option.
Fixes insufficient memory pressure on ZFS ARC by Virtual Machines to prevent out-of-memory conditions (NAS-135499).
Virtual machines created in 25.04 (pre-25.04.2) and displayed on the Containers screen do not automatically start on system boot to prevent conflicts with VMs on the Virtual Machines screen that might use the same zvol(s) (NAS-136946).
Makes changes to SMB shares:
Enables access-based enumeration for SMB shares with NFSv4 ACL type, so directory listings only show files that users have permission to read (NAS-136499).
Changes SMB share Advanced Options, Purpose, and Other Option settings associated with the preset setting selected (NAS-136499).
Removes the AUTORID IDMAP backend option from Active Directory configuration to improve consistency across multi-server environments (NAS-136630).
Existing configurations using AUTORID are automatically migrated to RID during upgrade.
Users should review their ACLs and permissions after upgrade and might need to reconfigure them in some edge cases.
Completes the transition to the versioned JSON-RPC 2.0 over WebSocket API by migrating all remaining jobs and events from the deprecated REST API (NAS-133984).
Full removal of the REST API is planned for the TrueNAS 26.04 release.
Removes the built-in SMART test scheduling and monitoring interface to improve user flexibility while maintaining smartmontools binaries for continued third-party script compatibility (NAS-135020).
Existing scheduled SMART tests are automatically migrated to cron tasks during upgrade, and users can install the Scrutiny app for advanced SMART monitoring.
SMART tests functions no longer show on the Data Protections Tasks, Storage Dashboard, or individual disk screens.
See Preparing to Upgrade for more information.
Improves drive temperature monitoring efficiency by extending the drivetemp kernel module to include SCSI/SAS disk temperatures.
Fixes an issue affecting drive temperature reporting on the dashboard (NAS-135572).
Fixes a bug to reenable available update notifications for custom apps (NAS-135124).
Fixes contrast issues on some UI theme options (NAS-135519).
In addition, we’re also beta testing a new format for TrueNAS Release Notes! Please take a look and send us your feedback - there’s a Feedback button on the right side of the release notes page for convenient reporting.
I’m shocked to see they have removed SMART UI options completely… I thought IX said they would improve it during a podcast and not just completely bin it? I’m honestly a bit shocked. I found this an integral part of TNS that was still severely dragging it’s feet but very important for the resilience of a storage server.
What’s going on here? Do we know what the intentions are? It says I should just run a third party app now if I’d want to keep track of my drives health… Isn’t such a thing supposed to be included in the UI of a storage server OS? and shouldn’t it be very robust and reliable instead of just ditching all and having to rely on third party solutions?
Removes the built-in SMART test scheduling and monitoring interface to improve user flexibility while maintaining smartmontools binaries for continued third-party script compatibility (NAS-135020).
Existing scheduled SMART tests are automatically migrated to cron tasks during upgrade, and users can install the Scrutiny app for advanced SMART monitoring
Sounds like a backwards step.
I’m assuming this means that you will no longer get smart warning/failure notifications?
It seems we will. It’s just very poorly explained in the blog post and here on the forums.
Overall it does feel like a minimal effort push. If you read the ‘Pros of removing SMART’, you really get the feeling that someone needed to write about why nobody wanted to allocate development time on it and dump the responsibility on some random third party app…
Honestly this should be included into the UI natively if you’d ask me.
You’ll need to test on real systems to make that assessment about the SMART plan. We’d like feedback on how well the new process works in real systems. We’re running it ourselves and so far its been good.
Here’s the summary of the rationale used to make the decision.
This epic will track the removal of the smartmontools integration in the UI and API.
The justifications for doing this are as follows:
our implementation was too strict and prevented community users from using smartd to monitor valid drives. (i.e. we “allowed” auxiliary parameters but we had a very strict block list which essentially made the auxiliary parameters worthless)
the scheduled smart tests section was buggy and very hard to maintain. While we have fixed numerous bugs for this over the years, the complexities kept growing and we continually encountered bugs
truenas stopped scheduling smart tests for SSD years ago because they don’t apply. SMART is a specification targeted for spinning platter disks. Flash based disks, while technically having “SMART INFO” available, is not a 1-1 mapping and give a false feeling of “security” in the sense of the drives being monitored.
when a SMART attribute for a disk reaches a certain threshold, it’s subjective on whether or not that indicates a “failed” or even a “failing” (predictive failure) disk. Our smart implementation would raise alerts for any type of “failure” that smartd implemented. This is not necessarily a bad thing but it can also produce false-positives which causes unnecessary support engagements.
PLEASE NOTE:
we are NOT removing smartmontools from TrueNAS base OS. Those binaries will be left in the product and they will continually be used by TrueNAS in specific use-cases. By leaving these binaries on the system, this also means any popular 3rd party scripts that utilize these tools will CONTINUE TO WORK. There should be no change on that front.
When someone upgrades from < GE release, any existing scheduled smart selftest tasks will migrate over and be put into the cron plugin and ran as a cron task.
Pros of “removing SMART”.
Increased flexibility for our community for monitoring disks that our implementation restricted.
A much better UX experience for SMART related drive statistics is provided by an app called “scrutiny”. This is in the truenas catalog and available to anyone that wants to install it. We have configured that app so that it automatically detects all drives on the host system and “just works” out of the box.
The scrutiny app also has the ability to store historical data in a database AS WELL AS be configured to send alerts based on certain conditions.
TrueNAS will still alert on certain SMART attributes for disks for everyone but the alerts will be targeted and will also allow us flexibility in the future to update (or add new) tests when the need arises.
(NOTE: We will alert on READ/WRITE/VERIFY uncorrected errors attributes for SCSI/SAS drives and on the “Reported Uncorrectable” attribute for ATA/SATA drives. NVMe doesn’t have an equivalent for this attribute. We will also alert on selftest failures. We will also check to see if any disk has run a selftest and if the last selftest has failed, we will raise an alert.)
We still allow someone to retrieve smartctl output from disks. This will be a UI feature. The idea is to return the raw text output of smartctl output.
An upstream kernel module “drivetemp” has been extended by TrueNAS to include SCSI/SAS disk temperatures. This allows us to very trivially and efficiently monitor drive temps instead of having to parse (potentially 1000’s of) files created by smartd.
===========================================
Personal View:
I think SMART was almost doubling the annual Failure Rate of drives through false positives. That is a lot of extra cost for system owners.
We have individual TrueNAS systems with over 1,000 drives in the field. Unnecessary drive failures are extra work and impact system performance with resilvering.
A known issue: Using ports other than 4420 and 8009 in ESXi can cause NVMe-oF to become undiscoverable. Similar issues may also occur with other software.
Another point to mention is that the current version of nvme-of uses the Linux kernel driver as the backend, so esxi cannot connect to the current truenas nvme-of.
The only thing the SMART functionality’s GUI does decently is let you schedule tests.
That’s it. The rest of it is practically useless.
Case in point - whenever someone has a potentially disk related issue the go-to suggestion is to run smartctl from the shell to get more stats. I honestly don’t think I have ever seen anyone suggest to go into the GUI to look at the SMART statistics - Perhaps that’s because it would just be a waste of time, there’s practically nothing there?
One can make the argument that the GUI should have been improved, I myself was hoping it would improve over the years. Alas, it has not, and the Jira comment gives a glimse of why that may be.
In the end I conclude that I will not be especially impacted by the removal of the GUI bits. Scheduling tests I want using cron or perhaps multi-report will handle my needs, and failed self tests and “Reported Uncorrectable” will still trigger a TrueNAS alert. Getting raw smartctl output in the UI is a long overdue addition that will be quite helpful when assisting people with their issues (consider including a Copy Raw Data-button for even more QoL!).
I noticed that the “Update Profile” feature seems to be missing the “Conservative” option. Without it, I still need to manually check the Software Status page to make sure I stay on a stable version. Since 25.04, I’ve decided not to update to anything below the Conservative profile (I even tested this in a VM first).
For that reason, the feature doesn’t feel very useful for my workflow right now. Do you have any thoughts on whether the Conservative setting might be added, or if there’s another way to handle this?
Thanks for the constructive review … we agreed that SMART support needed an overhaul we just wanted a sustainable model with fewer false alerts and more efficient disk temp stats.
A lot fewer potential software updates. The General setting will only be updated when software is considered mature and good quality.
A better Release Notes summary for each potential update.
We’ll review success of this base feature before trying to address the very “Conservative” users who, by definition, are the hardest group to test with. I’m personally a little worried by 100,000 conservative users updating on the same day. Murphy’s law would seem to apply.
My impression of Golden Eye was that SMART tests were going to get removed which as an end user I did not understand.
I see it is mentioned here it is a complex thing to keep work.
It looks like the functionality has been pushed into an App, if this is the new official way of doing things in GE, then nearer the time I think the TrueNAS team need to make people more aware of this.
You just sort of expect your appliance that uses hard disks/SSD/NVME to send you an email when a drive fails.
Maybe have scrutiny installed by default on new installs?
# Validation Error
7 validation errors for AlertClassesConfigResult result.classes.CatalogSyncFailed.policy Field required [type=missing, input_value={'level': 'NOTICE'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSBatteryLow.policy Field required [type=missing, input_value={'level': 'WARNING'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSReplbatt.policy Field required [type=missing, input_value={'level': 'WARNING'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSCommok.level Field required [type=missing, input_value={'policy': 'NEVER'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSCommbad.policy Field required [type=missing, input_value={'level': 'WARNING'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSOnBattery.policy Field required [type=missing, input_value={'level': 'WARNING'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing result.classes.UPSOnline.level Field required [type=missing, input_value={'policy': 'NEVER'}, input_type=dict] For further information visit https://errors.pydantic.dev/2.9/v/missing
Early Release Software
