Slimarr 1.7.0.0 Release Notes

Release date: 2026-06-23
Theme: Storage-Safe Automation and Persistent Jobs

This is a significant release focused on making NAS-backed libraries safe to
automate, giving long-running work a durable, recoverable home, and starting
a consistency pass on the UI. See CHANGELOG.md for the complete, dated
changelog across all releases; this document is the standalone summary for
1.7.0.0.

Headline fix: NAS freeze/crash during replacement

Some users on NAS-backed libraries reported the app appearing to freeze or
crash during file replacement. The root cause has been identified and fixed:

replace_file() — the function that swaps a downloaded file into your Plex
library — was calling shutil.disk_usage(), os.makedirs(),
os.path.exists()/getsize()/isdir(), and a recursive os.walk()
directly on the asyncio event loop instead of through a worker thread. On a
slow, busy, or sleeping NAS share, any one of those calls can block for
seconds. Because they ran on the event loop, that block stalled the entire
app — API requests, the websocket connection, and the scheduler — for the
duration. From the outside, that looks exactly like a freeze or crash.

Every blocking filesystem call in the replacement and duplicate-cleanup paths
now runs on a worker thread, matching the pattern already used elsewhere in
the NAS-safety code. Regression tests (tests/backend/test_replacer_event_loop.py)
assert the event loop keeps making progress while these calls are in flight,
rather than just trusting the call site.

Storage safety and persistent jobs

• A shared storage-safety layer (backend/core/storage.py) now handles path
  classification (local / NAS / network / recycling), preflight checks
  (existence, parent accessibility, free space), per-path locks, and optional
  NAS budgets: max write bytes/day, max replacements/day, max concurrent
  operations, and a cooldown after failures.
• Replacement, duplicate cleanup, failed-download cleanup, orphan cleanup,
  download-staging cleanup, and recycling-bin purges all route through this
  shared layer instead of raw shutil/os calls.
• Storage operation telemetry (recent moves/deletes, failures, durations,
  classifications) is now persisted to the database, surviving restarts, and
  exposed via Prometheus metrics, diagnostics bundles, and the System and
  Operations pages.
• A persistent job runtime (jobs / job_events tables) now backs manual
  scans, full automation cycles, duplicate preview/cleanup, and scheduled
  tasks, with heartbeats, stale-job recovery on restart, and cancel/retry
  APIs.
• Replacement recovery metadata is recorded before and after each risky
  filesystem phase (recycle original, backup existing target, place new
  file, restore on failure), so an interruption mid-replacement is visible
  and recoverable instead of silently leaving the library half-updated.
• A new Operations page (/system/operations) shows active and historical
  jobs, job event timelines, the storage operation log, NAS budget status,
  and guarded cancel/retry/purge actions.

UI consistency pass

• Unified destructive-action confirmation on one shared ConfirmDialog
  component: TV show delete (previously a one-off modal), blacklist entry
  removal, and orphaned-download cleanup (both previously fired immediately
  on click with no confirmation at all), plus the System page's recycling
  purge and duplicate cleanup (previously native window.confirm).
• Added shared Skeleton and EmptyState components and applied them
  consistently across Library, Queue, Operations, Orphaned Downloads, and
  Blacklist.
• Added a StorageStateMark component that maps replacement-recovery phase
  strings (preflight, recycling, backing up, placing, restoring, failed,
  recovery-required) to a consistent icon and color on the System page,
  instead of showing raw internal phase strings like place_replacement_started.
• Added a responsive mobile navigation drawer and reduced-motion support.

Implemented files.verify_after_download

This setting previously had no effect at all, despite defaulting to true
and printing a startup warning on every install ("not yet implemented").
It's now real: a downloaded file is rejected before it replaces a working
library copy if it's empty/zero-byte, and — when files.enable_media_probe
is also enabled — a warning is logged if no usable video stream is detected
(this does not hard-block, since probing can be flaky on some containers).

Security and code-health

• Removed several silent except Exception: pass blocks in the Plex
  integration and login-lockout audit path; failures are now logged.
• Added an identifier allowlist before table/column names are interpolated
  into SQLite migration DDL (defense-in-depth; these are hardcoded today,
  not exploitable, but the pattern is now safe if that ever changes).
• Added a startup warning when server.allowed_origins includes *. Risk
  is lower than typical wildcard-CORS setups because allow_credentials is
  False (no cookies sent cross-origin), but a leaked bearer token could
  still be used from any origin, so it's now surfaced instead of silent.
• Removed an accidentally committed 6MB log archive and a diagnostics
  history file from docs/logs-new/, and tightened .gitignore.

Docker / deployment documentation

docs/DOCKER.md previously only documented the pre-1.6.1 environment
variables and metrics. It now covers:

• All NAS-safety environment variables (SLIMARR_NAS_PATH_PREFIXES,
  SLIMARR_NAS_MAX_WRITE_GB_PER_DAY, SLIMARR_NAS_MAX_REPLACEMENTS_PER_DAY,
  SLIMARR_NAS_MAX_CONCURRENT_OPERATIONS, SLIMARR_NAS_FAILURE_COOLDOWN_MINUTES,
  SLIMARR_NAS_MAX_TRANSFER_MBPS, SLIMARR_NAS_COPY_CHUNK_MB,
  SLIMARR_MIN_SAVINGS_MB_FOR_NAS, SLIMARR_MIN_CYCLE_INTERVAL_MINUTES,
  SLIMARR_MAX_DOWNLOADS_PER_NIGHT, SLIMARR_THROTTLE_SECONDS,
  SLIMARR_MAX_ACTIVE_DOWNLOAD_HOURS).
• SLIMARR_DB_URL and PostgreSQL pool-tuning variables.
• The six Prometheus metrics added alongside persistent jobs and storage
  telemetry (slimarr_jobs_active, slimarr_jobs_failed_total,
  slimarr_nas_cooldown_active, slimarr_nas_storage_operations_active,
  slimarr_storage_operations_total, slimarr_storage_operation_bytes_total,
  slimarr_storage_operation_failures_total).
• A pointer to the System page's NAS Pressure panel and the non-mutating
  GET /api/v1/system/storage/preflight check.

.env.example, config.yaml.example, and all three Docker Compose variants
(docker-compose.yml, docker-compose.postgres.yml, docker-compose.traefik.yml)
already forwarded these settings correctly — only the prose documentation
was out of date.

Windows installer build fixes

While rebuilding the 1.7.0.0 installer, two real build-script bugs surfaced
and were fixed:

• build-installer.ps1 regenerated config.yaml.example from a hardcoded
  copy embedded in the script on every build, which had drifted from the
  real file — it was missing all NAS budget settings and set
  enable_media_probe: true (the unsafe default). The script now verifies
  the source-controlled file instead of overwriting it.
• The documented setup path (install.ps1) never actually installed
  pystray/pillow/pywin32, even though slimarr.spec requires them as
  PyInstaller hidden imports for the tray icon. Added requirements-tray.txt
  and wired it into install.ps1 on Windows.
• build-installer.ps1 only looked for "Inno Setup 6" in its installer
  detection; it now also finds Inno Setup 7.

Verification

• Backend: 93 tests pass on Python 3.12 (pytest tests/), including 12 new
  tests covering the event-loop fix, the SQL identifier guard, the
  verify-download logic, and the CORS wildcard warning.
• Frontend: tsc typecheck and vite build production build both pass
  cleanly with zero errors.

What's not in this release

The deeper visual redesign of Dashboard, Library, Movie Detail, and Settings,
plus new bitmap release/empty-state artwork (roadmap Phases 4-5 in
docs/VERSION_1_7_ROADMAP.md), are still in progress and will ship in a
follow-up release rather than being claimed here.

Docker image

The Dockerfile (python:3.12-slim runtime, multi-stage frontend build) was
reviewed for correctness but not built locally for this release — Docker
isn't installed in the environment this release was prepared in. Build and
push it the usual way before publishing:

docker build -t ghcr.io/theantipopau/slimarr:1.7.0.0 -t ghcr.io/theantipopau/slimarr:latest .
docker push ghcr.io/theantipopau/slimarr:1.7.0.0
docker push ghcr.io/theantipopau/slimarr:latest

Upgrade notes

• No manual migration steps. SQLite/PostgreSQL schema additions are additive
  and applied automatically on startup.
• If files.verify_after_download was already set to true in your config
  (the default), it now actually does something — see above. No action
  needed unless you want to disable it.
• If you rely on server.allowed_origins: ["*"], you'll see a new startup
  warning. It's informational; nothing is blocked.