WebDAV filesystem1.x
infinidysk exposes NZB contents as a browsable tree over WebDAV and the Explorer UI.
The WebDAV tree has four roots:
| Path | Role |
|---|---|
/content |
Mounted releases by category |
/nzbs |
Queued NZBs |
/completed-symlinks |
Symlink import entries (rclone --links turns them into symlinks) |
/.ids |
Stable ID-based paths that symlinks point to |
/view/{path} is not a WebDAV directory. It is a separate HTTP route that serves a file
from the tree for players, STRM files, and download links.
Content streams from Usenet on read; files are not downloaded to disk first. Blobs under
{CONFIG_PATH}/blobs/ store the NZB metadata needed to serve mounted files.
Credentials and read-only behavior: WebDAV settings. Mount with rclone for filesystem clients.
Explorer administration since 1.5.0
Section titled “Explorer administration since 1.5.0”Explorer opens the canonical /content hierarchy. Tree expands folders in place;
List shows matching files throughout the current scope. Expansion does not change the
URL or clear other branches, filters, selection, or an open player. Use the breadcrumb or
Use as scope to change scope explicitly. Existing /explore/content/... links still
open that directory; WebDAV system views retains access to /nzbs,
/completed-symlinks, and /.ids using their existing directory browser.
Search and metadata filters apply to the whole scope before pagination, including unopened folders. Both modes share health, stored schedule, library, category, indexer, storage kind, repair outcome, original NZB availability, size, and date filters. Search matches file names, paths, job names, and original NZB names. SQLite folds ASCII case in search; non-ASCII characters remain case-sensitive. Blank filters are unrestricted; Unknown is a distinct state. Each branch or list page contains at most 100 rows in the UI.
Reading file metadata
Section titled “Reading file metadata”- Health comes from the latest retained check for that exact file. Pending or urgent repairs take precedence. Unknown means there is no retained result, even if a last-check timestamp remains after history cleanup. Requeueing a degraded file does not make it healthy.
- Next scan distinguishes checking, disabled, repair pending, queued, due, and scheduled work. Queued sentinels are never shown as dates. No scheduled date filters the stored null date, so it can include unsupported sidecars. A due date is eligibility, not a promised start time: health windows, queue pressure, and workers can defer execution.
- Added age uses the existing server-local added timestamp, serialized with an offset. Posted age uses the release date and remains unknown when that date is absent. They are not interchangeable. Historical added dates cannot recover a previous server timezone; negative displayed ages are clamped to zero.
- The focused row’s details include canonical path and ID, storage kind (NZB, RAR, or multipart), health outcome and message, job, original NZB name, category, indexer, and library link paths. Last played (release) belongs to the linked release history, not an individual file. Deleting history does not hide a live file.
- Library membership is a completed scan of recognized symlinks or STRM files in the configured Library Directory, matched by exact file ID. It is refreshed periodically and is not a live Arr inventory. Pending or failed scans mean Unknown, never Not in library. No configured root means Not configured. The scan timestamp or failure is shown with the file count. Link details contain filesystem paths, not signed STRM target URLs.
File actions
Section titled “File actions”Recheck queues only the selected supported file and preserves its observed health and repair history. It respects disabled repairs, work windows, queue pressure, and worker concurrency. Repeating a request for an already queued or running file does not start duplicate work. WebDAV enforce-read-only blocks removal, not an administrator’s recheck.
Play uses the existing built-in player when the media type is supported. Download streams the file through its signed URL. Export NZB uses the retained original NZB identity and filename; an unavailable original is not reconstructed. Media inspection remains on demand in the player.
Search in Arr is a confirmed, non-destructive request to enabled Radarr/Sonarr instances that own the file. It requires a recognized link in the configured Library Directory and matching ownership paths; a download ID alone is not enough. Every enabled instance must be checked before any search is requested. infinidysk does not remove or blocklist the current file, alter monitoring or quality rules, or spend the automatic replacement-search budget. Arr’s quality, cutoff, and monitoring decisions still apply, so a search receipt does not guarantee a replacement download. Multiple owners can produce partial results. Unconfirmed means the command might have arrived; check Arr before manually trying again. These requests are not automatically retried.
Read-only frontend accounts can browse, play, download, and export but cannot recheck, search, preview removal, or remove. Disabled action tooltips explain the restriction. Mobile layouts retain every filter and action through toolbar wrapping and horizontal table scrolling.
Existing Library Directory, Arr instance, Repairs, and WebDAV settings govern these actions.
Deleting from Files since 1.1.0
Section titled “Deleting from Files since 1.1.0”From Explorer, you can delete individual files or entire release folders under /content/{category}/{release}/…. Removal requires a completed impact preview before confirmation. The dialog shows file, folder, byte, and linked-history totals.
Selecting a directory includes its current collapsed, hidden, and filtered-out descendants. Overlapping parent/child selections are reduced to their top-level targets. Counts are advisory at preview time: the dialog does not lock a subtree against new arrivals. Bulk removal runs sequentially, is not atomic across directories, and retains individual failures for review without retrying already removed targets. The selected file ID is checked again, so a different file appearing at the same path cannot be removed using a stale selection.
Protected roots and category folders, matching active downloads, and Enforce Read-Only continue to prevent removal.
Deletion is limited to mounted content under /content. Use the Queue page to remove NZBs from /nzbs, and the History page to clear /completed-symlinks entries. Internal paths such as /.ids cannot be deleted from Files.
When the last file referencing a history entry is removed, the history row is pruned automatically so external SAB clients do not keep pointing at deleted mounts.