SABnzbd1.x

SABnzbd-compatible download client API used by Radarr/Sonarr. See also API compatibility.

Setting	What it does
API Key `api.key`	Arr download client auth. Default from `FRONTEND_BACKEND_API_KEY` if unset.
Categories `api.categories`	Letters, numbers, and dashes. Default env or `audio,software,tv,movies`.
Manual Upload Category `api.manual-category`	Category for NZBs uploaded from the Queue page; also the default of the category picker beside the Upload NZB button since 1.2.0. Default `uncategorized`.
Trusted local hosts since 0.8.0 `api.addurl-trusted-hosts`	Hostnames, IPs, or CIDR ranges `addurl` may fetch from even though they resolve to private or loopback addresses. Default env `TRUSTED_INTERNAL_HOSTS`.
Import Strategy `api.import-strategy`	Symlinks (Plex) or STRM Files (Emby/Jellyfin). Default `symlinks`.
Rclone Mount Directory `rclone.mount-dir`	Shown for Symlinks. Path where rclone mounts the WebDAV root. Default env `MOUNT_DIR` or `/mnt/nzbdav`.
Completed Downloads Dir `api.completed-downloads-dir`	Shown for STRM Files. Directory visible to Arr apps where STRM files are written. No default; suggested `/data/completed-downloads`.
Ignored Files `api.download-file-blocklist`	Wildcard patterns (`` and `?`) for files that are not mounted. Default `.nfo, .par2, .sfv, unpack.mkv, .unpack.mp4`.
Filter sample videos from downloads since 0.10.0 `api.sample-filter-enabled`	Discard videos in sample release subfolders such as `Sample/` since 1.3.0, even when no main video exists. Filename-only whole-word `sample`/`samples` matches are discarded only when under 20% of the largest video in the NZB. Category and release folder names are ignored. Default on.
Rename a single video to the release name since 1.3.0 `api.rename-single-video-to-release`	When a job mounts exactly one video, rename it to `{release-folder}{extension}`. Season packs and collisions keep the original name. New imports only. Default on.
Behavior for Duplicate NZBs `api.duplicate-nzb-behavior`	`increment` downloads again with a suffix; `mark-failed` fails the download. Default `increment`.
Fail downloads for NZBs without video or audio content `api.ensure-importable-video`	Fail NZBs with no media after filtering; failures identify RAR-inside-7z limitations when applicable. Turning this off permits raw archive mounts, not nested-archive playback. Default on.
Fail downloads when non-media files have missing articles inverse of `api.skip-non-video-on-missing-articles`	Video and audio files always fail on missing articles; other files are skipped unless this is on. Default off.
Perform article health check during downloads since 1.2.5 `api.ensure-article-existence-categories`	Categories that get STAT checks plus a final direct-media BODY readiness read; may be slow. Default empty (off).
Article health check mode `api.article-existence-check-mode`	Full or per-file sampled verification. Default `full`.
Always send full History to Radarr/Sonarr `api.ignore-history-limit`	Ignore the client’s History limit. Default on.
Save backup copies of incoming NZBs `api.nzb-backup-enabled`	Write a copy of each incoming NZB to disk. Default off.
Backup directory `api.nzb-backup-location`	Where backups go, in category subdirectories. Default empty.
Keep NZB backups for (days) `api.nzb-backup-retention-days`	`0` keeps them forever. Default `30`.

STRM files also need the public Base URL, which is under Settings → General → Public access, not this tab; see URL base.

Queue capacity and admission limits are configured separately under Queue. The default user agent for retrieving NZBs, including matched addurl requests, is configured under Indexers.

STRM sidecar cleanup since 1.3.0

Generated STRM files are deleted together with their content (history delete-with-files, health repair, and the Remove Orphaned Files maintenance task). Sidecars under the configured completed-downloads directory never count as library links for unlinked-file cleanup, even when that directory sits inside the Library Directory.

Sampled article checks since 0.10.0

Categories selected under api.ensure-article-existence-categories use a full article check by default, preserving existing behavior. Set api.article-existence-check-mode to sampled to check the first and last segments plus an evenly spaced selection of middle segments in each important file. Sampling is applied per file so every file’s tail is covered; this catches common truncated or partially removed releases without a full STAT sweep.

Small files (currently up to roughly 8,000 segments) are still checked in full. The sampled mode uses the same standard-depth selection as background health checks. It is a screen rather than a guarantee: urgent repair remains the backstop if an article disappears later or STAT succeeds while BODY fails.

For selected categories, infinidysk also reads the head and tail of direct media outputs before reporting a successful SAB completion since 1.2.5. This rejects unreadable container headers and short reads before Sonarr or Radarr begin import. It does not replace streaming corruption detection or PAR2 repair: damage limited to the middle of a release can only be detected when that part is read.

API-key rotation since 1.2.5

After rotating api.key, update the API key in every configured Sonarr/Radarr download-client entry and use its test action before resuming a bulk search. Rejected SAB API requests are logged without exposing the key, with failures for the same mode and category throttled.

Import strategies · Queue settings