Hatsune Imagine f5956999a7 feat(webdav): report quota properties on PROPFIND (#8197) ...

dav-server queries get_quota once per PROPFIND without a path, so the
driver now reports a single session-scoped (used, total) pair:

- when every bucket visible to the session has a hard quota: the summed
  cached bucket usage against the summed quota limits (a bucket whose
  usage cache has no scanner snapshot yet counts as zero);
- otherwise: the cluster's usable capacity, the same erasure-aware
  used/total the console dashboard shows, cached for 30s since WebDAV
  clients poll quota and storage-info fans out to every disk;
- when neither source is available the properties are omitted from the
  response, which is the previous behavior.

All inputs are cache or snapshot reads; reporting never triggers a
scanner cycle or a live listing. Backend failures surface as
FsError::GeneralFailure instead of a bogus zero capacity.

Adds a session_capacity_view hook to the protocols StorageBackend trait
(default: no capacity support, keeping quota properties omitted), its
rustfs-side implementation, aggregation unit tests, and an e2e test
covering both reporting modes.

2026-10-03 16:50:24 +08:00

..

src

feat(webdav): report quota properties on PROPFIND (#8197)

2026-10-03 16:50:24 +08:00

tests

chore(deps): migrate direct encoding deps to simd (#6690)

2026-08-27 03:17:24 +08:00

Cargo.toml

chore(deps): refresh RustFS tokio tar dependency (#7907)

2026-09-15 14:54:21 +08:00