Understanding ZFS Scrubs and Data Integrity

ZFS “scrubs” explained — but commenters say “how often?” and “it’s not a magic fix”

TLDR: ZFS scrubs are routine scans that verify every block of data to catch silent corruption, and the article says they’re essential. Commenters debated how often to run them, warned scrubs don’t fix messy files, and argued over error-rate math—useful because your backup habits may hinge on these details.

The article paints ZFS scrubs as a full-body checkup for your storage, and the crowd largely agrees: ZFS is the “trust-no-one” filesystem that re-checks every block with a checksum and fixes bad ones with redundancy. But the comments stole the show. The top question, from itchingsphynx, was pure homeowner energy: how often do you run this thing? Monthly is common, sure—but if your data changes fast, is that enough? Cue a flood of “depends on churn” replies, coffee-and-cron memes, and the classic “not that kind of scrub” jokes.

Then came the big reality check from thatcks: scrubs don’t tidy up everything. They verify that bytes match their fingerprint, but they don’t fix broken file structures like older tools (fsck/chkdsk) try to do. Translation: great smoke detector, not a carpenter. That line sparked a mini panic and a chorus of “so what else should I run?”—with veterans reminding newbies that scrubs ≠ repairs for messy directories.

And because it’s the internet, the math fight arrived right on schedule. klempner challenged the scary hard-drive error-rate stat, and suddenly we had charts, napkin math, and “actually”s flying everywhere. Also spotted: gentle eye-rolls at the vendor-y sales pitch, and one gem of wisdom—“Set it and schedule it, but don’t worship it.” In short: scrubs are vital, not mythical.

Key Points

  • ZFS scrubs verify every allocated block of data and metadata, including parity, against stored checksums.
  • ZFS uses a Merkle tree: parent block pointers contain child checksums, ensuring end-to-end integrity and exposing corruption.
  • During scrubs, ZFS recalculates checksums and repairs mismatches using redundancy (e.g., mirrors).
  • ZFS scrubs differ from tools like fsck/chkdsk; ZFS’s transactional design maintains metadata consistency without logical repairs during scrubs.
  • HDDs have typical bit error rates around 1 in 10^15, making regular verification important to detect silent corruption; scrubs are not the same as resilvers.

Hottest takes

"Is there a more specific 'rule of thumb' for scrub frequency?" — itchingsphynx
"ZFS scrubs do not check filesystem objects for correctness and consistency" — thatcks
"While the advice is sound, this number isn't the right number for this" — klempner
Made with <3 by @siedrix and @shesho from CDMX. Powered by Forge&Hive.
Understanding ZFS Scrubs and Data Integrity - Weaving News | Weaving News