v0.6.363: account policy (one-time admin resets, 15-character passwords, password-changed audit) #1029

Merged
Cordy merged 11 commits from r6 into main 2026-10-27 17:53:08 +00:00
Owner

Account policy from the #904 decisions (2026-10-27):

  • One-time admin resets: every admin reset of a local account is a one-time password (16 generated characters, valid 7 days). The person chooses their own at the next sign-in (frame K). Rows show "New password pending" or "One-time password expired", and the person gets a bell item.
  • Password length: new local passwords need at least 15 characters, counted after NFC, on every path including cairnd hash-password. Existing passwords keep working.
  • Same password refused: at the new-password step and at Change password (password-same).
  • Audit: a new password-changed entry; user-password-reset one-time=yes.
  • Standards: security-standards.md checked against NIST SP 800-63B-4 and OWASP ASVS 5.0 primary texts.

Review: backend (1 blocker, a race at the step, fixed with a generation-checked write) and UI each reviewed, plus a whole-branch review: one should-fix (expired one-time password at Change password), fixed.

Part of #904; the restore declaration itself follows in v0.6.364.

Account policy from the #904 decisions (2026-10-27): - **One-time admin resets:** every admin reset of a local account is a one-time password (16 generated characters, valid 7 days). The person chooses their own at the next sign-in (frame K). Rows show "New password pending" or "One-time password expired", and the person gets a bell item. - **Password length:** new local passwords need at least 15 characters, counted after NFC, on every path including `cairnd hash-password`. Existing passwords keep working. - **Same password refused:** at the new-password step and at Change password (`password-same`). - **Audit:** a new `password-changed` entry; `user-password-reset one-time=yes`. - **Standards:** security-standards.md checked against NIST SP 800-63B-4 and OWASP ASVS 5.0 primary texts. **Review:** backend (1 blocker, a race at the step, fixed with a generation-checked write) and UI each reviewed, plus a whole-branch review: one should-fix (expired one-time password at Change password), fixed. Part of #904; the restore declaration itself follows in v0.6.364.
Two restores of one deleted space on two instances (no shared lock): B
moved the folder, A finished the restore (revived, removed the entry and
every intent), B's revive then answered space-gone because Undelete reads
the record as written since #1007 (v0.6.356) and found it live, and B's
rollback check read the entry (still there) before the intents (already
gone, so none foreign): B moved the live space's folder back into the
trash. CI flake of TestSpaceRestoreRestoreConcurrent935/s3.

- Revive: a tombstone gone from the record as written is answered as a
  missing one (nil when the space is live on its folder).
- mayRollBack reads the intents before the entry: a finishing restore
  removes the entry, then the intents, so a finish in between shows.

Deterministic tests on posix and s3: TestSpaceRestoreReviveAfterAnotherRestore935,
TestSpaceRestoreNoRollbackAfterAnotherFinished935.
docs(v0.6.363): CHANGELOG fix line for concurrent space restores
All checks were successful
ci / test-and-build (pull_request) Successful in 11m48s
6ea663d14b
Cordy merged commit 3b7a9c8905 into main 2026-10-27 17:53:08 +00:00
Sign in to join this conversation.
No reviewers
No labels
No milestone
No project
No assignees
1 participant
Notifications
Due date
The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference: Cordy/Cairn#1029
No description provided.