An interactive museum that reads ZFS in its source code, focusing first on the uberblock - the tiny record that names an entire pool. Every animation and excerpt links to exact lines from OpenZFS zfs-2.4.4 (commit 71a9f9...), so the presentations are literal mappings to code. The hall explains how the uberblock holds ub_txg and ub_rootbp, how each disk label stores a 128 KiB ring of uberblocks, and how new uberblocks rotate into slots using txg mod slot count (with multihost/MMP behavior noted). Concrete specifics include slot sizing tied to sector size/ashift (128 slots on 512B sectors, fewer on larger sectors), the import rule of choosing the valid uberblock with highest txg, and the shared magic 0x00bab10c that begins every uberblock.
The site then routes visitors through rooms that unpack ZFS mechanisms: the root and tree, copy-on-write, checksums and self-healing, transaction groups, vdev layouts (mirror, raidz, draid), RAID-Z stripe math, snapshots and clones, ARC caching, metaslabs and space maps, ZIL/SLOG, send/receive, and write-pipeline transforms (compression, dedup, encryption). It frames these explanations as an homage to ZFS engineering and history, cites sources and authors, and documents how the museum itself was built, with code licensed under CDDL and explanatory text CC BY.
Summary generated by AI from the linked article. hn.today is not affiliated with Hacker News or Y Combinator.