Object storage has quietly become the duct tape of self-hosting. Backup tools want an S3 endpoint. Photo libraries and media servers keep asking where the S3 bucket is. Even Proxmox Backup Server now accepts S3 as a target. For years, the homelab answer was the same: spin up MinIO in a container and point everything at it. That answer just got more complicated, and TrueNAS picked an interesting moment to offer a new one.
TrueNAS 27 RC.1, the first release candidate of iXsystems next major NAS operating system, ships an early-access feature called the TrueNAS Object Interface: native S3-compatible storage built directly into the platform. No container, no third-party software. An S3 bucket maps straight to a ZFS dataset. The release candidate is available for testing now, with the final 27.0 release expected in early December 2026.
What the Object Interface actually does
The concept is straightforward. Today, getting S3-compatible storage on a NAS means running something like MinIO or Garage in a Docker container, managing its own disk allocation, its own configuration, and its own update cycle, separate from the storage pool it sits on. The Object Interface collapses that layer. When you create a bucket, it is a ZFS dataset. That means it inherits everything ZFS already gives you: snapshots, quotas, replication, and compression, with no extra integration work.
The initial early-access release supports the core S3 feature set: multipart uploads, object tagging, bucket and object ACLs, versioning, and immutability. It is available across TrueNAS editions with no capacity limits, and it can run in offline, air-gapped environments, which matters for homelab users who run their storage network on an isolated VLAN. Two features sit behind a paywall: versioning and Object Lock require TrueNAS Connect Plus or TrueNAS Enterprise. iXsystems is targeting full production status for the interface in 2027, so this is a preview, not a finished product.
Why the timing matters: the MinIO license squeeze
Enjoying this story?
Get the five most important stories in tech, every morning. Free.
To understand why a native S3 interface lands so loudly, look at what happened to the default option. MinIO, long the go-to self-hosted S3 server, has changed its licensing terms in a way that homelab writers say has effectively gutted the free, open source version as a practical choice. The specifics of the licensing shift are MinIO business to explain, but the community reaction is visible: forum threads and self-hosting bloggers are actively shopping for replacements.
A bucket that is a ZFS dataset inherits snapshots, quotas, and replication for free, which is exactly the combination backup tools want from an S3 target.
TrueNAS is positioning itself as that replacement. Because the Object Interface lives on every TrueNAS edition without capacity limits, a home user with a four-bay NAS gets the same bucket mechanics as an enterprise customer. The trade-off is maturity. MinIO is a battle-tested, decade-old object server. The TrueNAS Object Interface is an early-access feature aimed at production readiness sometime in 2027. Anyone pointing real backup jobs at it should treat the RC period as a test lab.
The rest of RC.1: LXC graduates, search arrives
TrueNAS 27 RC.1: What is New
The homelab-relevant highlights, ranked by impact for self-hosters.
Impact scores are the author editorial assessment for typical homelab use. The S3 interface is early access; production status is targeted for 2027.
The S3 interface is the headline, but the release candidate carries two other changes that matter for home labs. First, LXC containers are now fully supported, graduating from experimental status in earlier releases. Proxmox users already know LXC as the lightweight way to run Linux services; TrueNAS users can now run small services directly on their NAS with official support. There is a networking caveat worth knowing: TrueNAS documents an automatic bridge option, and MACVLAN on the same physical interface cannot communicate with the host itself, so container networking needs planning.
Second, TrueNAS 27 adds TrueSearch and WebShare. TrueSearch brings server-side indexing of file names, metadata, and file contents, including macOS Spotlight integration so Finder can search indexed NAS shares without the client crawling the network. WebShare adds browser-based file access from any device, with existing ACLs and permissions intact. Both require TrueNAS Connect, and encrypted datasets are excluded from indexing.
Underneath, the release moves to OpenZFS 2.4 and Linux kernel 6.18. OpenZFS 2.4 brings deduplication writes up to 25 percent faster, per TrueNAS numbers, faster encryption on CPUs with AVX2, default user, group, and project quotas, and the ability to restrict scrub operations to defined time windows. The kernel bump adds hardware support for Broadcom tri-mode HBAs, PCIe Gen5 switching, and newer NVIDIA GPUs.
What should you do before December

If you run TrueNAS in production, the official advice is unchanged: release candidates are for testing, and production systems should wait for 27.0 in early December. If you are a homelab user affected by the MinIO licensing changes, the practical path is to spin up RC.1 in a test VM, create a bucket, and point a non-critical workload at it, like an offsite backup copy or a photo-sync target. Test the failure modes: snapshots of a bucket dataset, restores, and what happens when you exceed a quota.
Watch the licensing of the advanced features, too. Versioning and Object Lock requiring Connect Plus or Enterprise is a deliberate choice, and immutability without versioning has real limits for ransomware protection. For users whose backup strategy depends on object locking, that distinction could decide whether the Object Interface replaces MinIO or merely supplements it.
The bigger story is the trend line. TrueNAS is absorbing functions that used to need separate containers: first apps, then containers, now object storage. The NAS is becoming the platform, not just the disk shelf. For self-hosters, that is one fewer thing to maintain, which, as anyone who has patched four services in a weekend knows, is the whole point.
0 Comments