Current state.. Here’s the current evidence-backed feature list. This is real upstream OpenZFS running on AROS—not the earlier replacement filesystem. “Verified” below means exercised in isolated AROS hosted/QEMU tests; it does not mean certified for your physical 50 TB array. Filesystem and AROS integration — implemented and verified • Native DOS-mounted datasets, accessible through Wanderer. • File creation, reading, writing, append, seek, resize, flush and close. • Directory creation, enumeration, deletion, rename and moves. • Hard links, symbolic links and file/lock identity handling. • DOS64 sparse-file operations. • Persistent dates, comments, ownership and protection. • UTF-8 filename handling, normalization and case-insensitive lookup. • Wanderer drawer updates following filesystem changes. • Read-only enforcement and busy-handle protection during unmount/export. • Upstream ACL evaluation integrated with security.library/MuFS identities. • Selected intent-log replay and recovery cases, including failed-replay cleanup. These have native execution and persistence tests—not just compilation. [Filesystem evidence](/home/terminills/Documents/GitHub/openzfs-aros/aros/zpl/README.md), [Wanderer verification](/home/terminills/Documents/GitHub/openzfs-aros/aros/zpl/WANDERER-UNLINK.md) Pools, disks and RAID — implemented and verified • File-backed pool creation: stripe, mirror, RAIDZ1, RAIDZ2 and RAIDZ3. • Native device/partition discovery, geometry queries and GPT operations. • Native partition-backed mirror import, writes, export and read-only reimport. • Native partition-backed RAIDZ1: healthy and missing-member operation, with subsequent cold-boot verification. • Mirror survivor-only recovery. • File-backed expansion, attach, replacement and completed resilver. • Populated mirror splitting, with independent cold reads from both resulting pools. • Device evacuation/removal, followed by deleting the removed backing file and reading the surviving pool. • Scrub, reopen, online/offline, detach, pool checkpoints and GUID changes. • Pool status, capacity and cumulative I/O statistics. Coverage differs by interface and backing type: attach/replace/split have native management-API tests; that does not establish every equivalent CLI or physical-device workflow. [Pool evidence](/home/terminills/Documents/GitHub/openzfs-aros/aros/engine/README.md), [Native RAIDZ evidence](/home/terminills/Documents/GitHub/openzfs-aros/aros/engine/NATIVE-RAIDZ-VERIFICATION.md) Datasets and data services — implemented and verified • Dataset creation, enumeration, properties, rename and destruction. • Snapshots, recursive snapshot rename, clones, promotion and rollback. • Snapshot holds/releases and bookmarks. • Compression and selected space-accounting checks. • Full and incremental send/receive through the native API. • Resumable receive, token-based continuation and interrupted-receive abort. • Selected compressed, large-block and embedded stream representations. • Upstream encryption with raw, hexadecimal and passphrase key files. • Key load/unload/change, wrong-key refusal, mounted-key protection and cold encrypted-data readback. These are selected verified workflows, not every OpenZFS option combination. [Replication evidence](/home/terminills/Documents/GitHub/openzfs-aros/aros/engine/RESUMABLE-RECEIVE.md), [Encryption evidence](/home/terminills/Documents/GitHub/openzfs-aros/aros/engine/ENCRYPTION.md) Actual upstream commands — verified on AROS The real

zfs

and

zpool

executables link and run against the resident service. Verified command workflows include: • Pool list/status, discovery, import/export. • File-backed create, add and destroy. • Sync, scrub, reopen, non-rewind clear and checkpoints. • Online/offline, mirror detach, removal and removal dry-run. • Pool GUID changes. • Dataset create/list/get/set/mount/unmount/rename/destroy. • Snapshot creation, listing, rename and destruction. • Snapshot space accounting. • Permission and read-only refusals across the tested operations. [CLI verification and exact limitations](/home/terminills/Documents/GitHub/openzfs-aros/aros/cli/README.md) Native Manager and startup — implemented and verified • Real authenticated service requests, results and error handling. • Interactive verification of selected management, encryption, replication and statistics workflows. • Validated IFF preferences, with `ENV:`/`ENVARC:` handling. • Protected Save for boot records. • Expected-pool-GUID validation before startup import. • Pool-scoped startup mounting that respects mount policy. • Two separate QEMU boots: save persistent settings, then restore, import, mount and read the pool in the fresh guest. Production startup installation is still unfinished; this is verified private boot testing, not an installed production configuration. Not finished / not yet proven • Native-partition CLI

zpool add

: current work; latest test failed, so not counted as verified. • Complete CLI feature/option parity. • Whole-device provisioning and all auxiliary-vdev/allocation-class workflows. • Actual device TRIM/discard. • NFS/SMB sharing providers. • Secure interactive key prompting/encrypted key storage. • Final standalone handler/shared-engine packaging and stock-nightly compatibility. • Large-array scalability, sustained stress, comprehensive fault injection and physical power-loss/flush correctness.