Files
Deluan Quintão cc315dcc8c perf(db): keep query planner statistics trustworthy with full ANALYZE (#5740)
* perf(db): keep query planner statistics trustworthy with full ANALYZE

PRAGMA optimize's internal ANALYZE runs with a limited analysis budget
(~2000 rows) that writes wrong sqlite_stat1 entries for low-cardinality
indexes: on a 96K-track library it claimed (missing, library_id) narrows
to ~2000 rows when it matches the whole table. The planner then prefers
that index over the sort index and falls back to a full-table temp
B-tree sort per request, turning paginated song listings into
multi-second queries (reproduced at 5.5s on real hardware; ~90x slower
than with correct stats). Every index-creating migration re-triggered
the poisoning via the post-migration optimize, and the daily optimizer
could re-trigger it on large library changes. Setting analysis_limit on
the connection does not help: optimize ignores it.

Run a plain full ANALYZE instead: after migrations with schema changes,
and in db.Optimize (daily schedule and scan-end). Stats are stored in
the database file, so one connection suffices and the per-connection
pool loop is gone. The Optimize call at shutdown is removed: stats are
maintained at migration/scan/daily points, and an ANALYZE during
shutdown only delays it and races container stop timeouts.

* perf(db): drop startup PRAGMA optimize that re-poisons planner stats

The startup PRAGMA optimize=0x10002 runs SQLite's budget-limited internal
ANALYZE (bit 0x02), which writes truncated sqlite_stat1 rows for
low-cardinality indexes -- the exact statistics-poisoning this PR set out to
eliminate. Because DevOptimizeDB defaults to true, a restart with no pending
migrations would re-poison the planner until the next scan or daily Optimize.

Remove it: statistics are already refreshed with a full ANALYZE after
schema-changing migrations (Init) and via Optimize at scan-end and on the daily
schedule, so nothing on the startup path needs to touch them. Also clarify that
Optimize is a no-op unless DevOptimizeDB is enabled.

* chore(db): remove the DevOptimizeDB flag and skip Optimize on quick scans

The flag only gated the optimize/ANALYZE maintenance calls and there is
no reason to leave planner statistics unmaintained; the guards are gone
along with the flag. The scan-end Optimize now runs only after full
scans — quick scans barely move the statistics, and the daily schedule
covers drift.

* style(scanner): drop redundant comment in runOptimize

* chore(persistence): drop the no-op PRAGMA optimize from ScanEnd

Mask 0x10000 only selects candidate tables by size change; without the
0x02 action bit optimize does nothing (verified: sqlite_stat1 stays
stale after a 100x table growth). The scan-end statistics refresh is
db.Optimize's full ANALYZE, and the expression-collation-index concern
the old comment guarded against no longer applies.

* fix(scanner): run the post-scan ANALYZE in the server process

With the external scanner (the default), the scan pipeline runs in a
subprocess, so its ANALYZE was invisible to the server: SQLite loads
sqlite_stat1 into the process's shared schema cache, and an ANALYZE from
another process does not refresh it — verified with the production DSN
that even brand-new pool connections keep planning with the old
statistics until the server restarts. An in-process ANALYZE, by
contrast, is immediately visible to every pooled connection through the
same shared cache.

Move the full-scan Optimize from the scanner pipeline to the scan
controller, which always runs in the server process.

* fix(scanner): honor promoted full scans in the optimize gate

A quick scan resuming an interrupted full scan is promoted inside the
scanner (possibly in a subprocess); mirror the promotion in the
controller so the post-scan ANALYZE isn't skipped.

* refactor: apply cleanup review findings

- drop forceFullRescan's inline ANALYZE: Init already runs a full
  ANALYZE after any migration batch with schema changes, so upgrades
  including a full-rescan migration analyzed the whole DB twice
- resumingFullScan uses a filtered CountAll instead of fetching and
  scanning all libraries
- document why CallScan (CLI) deliberately skips the post-scan Optimize

* perf(db): make planner analysis maintenance resilient

Check analysis freshness every 30 minutes and refresh statistics when the last successful run is over 24 hours old or a scan marked them pending. Persist successful analysis state, retry skipped or failed maintenance, coordinate checks with scans, and cover standalone CLI full scans.

* perf(db): avoid analyzing routine quick-scan changes

Reserve pending analysis for full scans, unscanned libraries, and retry state. Incremental quick scans now rely on the 24-hour freshness window instead of triggering a full ANALYZE at the next maintenance check.

* fix(scan): analyze resumed full scans in CLI

* fix(db): back off failed analysis retries

* feat(db): allow disabling scheduled analysis

* test(db): remove redundant analysis coverage

* refactor(db): split ANALYZE maintenance into optimize.go and dedupe call sites

- move query-planner statistics code from db.go to its own optimize.go
  (and matching optimize_test.go)
- log ANALYZE elapsed time inside Optimize/OptimizeIfNeeded instead of
  repeating the timing block at every call site
- drop the LastDBAnalyzeAttemptAt write on success: it is only read
  while failures >= 1, and every failure rewrites it first
- extract runPostScanAnalysis (cmd) and anyIncludedLibrary (scanner)
  helpers
2026-07-13 12:04:29 -04:00
..

Navidrome Scanner: Technical Overview

This document provides a comprehensive technical explanation of Navidrome's music library scanner system.

Architecture Overview

The Navidrome scanner is built on a multi-phase pipeline architecture designed for efficient processing of music files. It systematically traverses file system directories, processes metadata, and maintains a database representation of the music library. A key performance feature is that some phases run sequentially while others execute in parallel.

flowchart TD
    subgraph "Scanner Execution Flow"
        Controller[Scanner Controller] --> Scanner[Scanner Implementation]
        
        Scanner --> Phase1[Phase 1: Folders Scan]
        Phase1 --> Phase2[Phase 2: Missing Tracks]
        
        Phase2 --> ParallelPhases
        
        subgraph ParallelPhases["Parallel Execution"]
            Phase3[Phase 3: Refresh Albums]
            Phase4[Phase 4: Playlist Import]
        end
        
        ParallelPhases --> FinalSteps[Final Steps: GC + Stats]
    end
    
    %% Triggers that can initiate a scan
    FileChanges[File System Changes] -->|Detected by| Watcher[Filesystem Watcher]
    Watcher -->|Triggers| Controller
    
    ScheduledJob[Scheduled Job] -->|Based on Scanner.Schedule| Controller
    ServerStartup[Server Startup] -->|If Scanner.ScanOnStartup=true| Controller
    ManualTrigger[Manual Scan via UI/API] -->|Admin user action| Controller
    CLICommand[Command Line: navidrome scan] -->|Direct invocation| Controller
    PIDChange[PID Configuration Change] -->|Forces full scan| Controller
    DBMigration[Database Migration] -->|May require full scan| Controller
    
    Scanner -.->|Alternative| External[External Scanner Process]

The execution flow shows that Phases 1 and 2 run sequentially, while Phases 3 and 4 execute in parallel to maximize performance before the final processing steps.

Core Components

Scanner Controller (controller.go)

This is the entry point for all scanning operations. It provides:

  • Public API for initiating scans and checking scan status
  • Event broadcasting to notify clients about scan progress
  • Serialization of scan operations (prevents concurrent scans)
  • Progress tracking and monitoring
  • Error collection and reporting
type Scanner interface {
    // ScanAll starts a full scan of the music library. This is a blocking operation.
    ScanAll(ctx context.Context, fullScan bool) (warnings []string, err error)
    Status(context.Context) (*StatusInfo, error)
}

Scanner Implementation (scanner.go)

The primary implementation that orchestrates the four-phase scanning pipeline. Each phase follows the Phase interface pattern:

type phase[T any] interface {
    producer() ppl.Producer[T]
    stages() []ppl.Stage[T]
    finalize(error) error
    description() string
}

This design enables:

  • Type-safe pipeline construction with generics
  • Modular phase implementation
  • Separation of concerns
  • Easy measurement of performance

External Scanner (external.go)

The External Scanner is a specialized implementation that offloads the scanning process to a separate subprocess. This is specifically designed to address memory management challenges in long-running Navidrome instances.

// scannerExternal is a scanner that runs an external process to do the scanning. It is used to avoid
// memory leaks or retention in the main process, as the scanner can consume a lot of memory. The
// external process will be spawned with the same executable as the current process, and will run
// the "scan" command with the "--subprocess" flag.
//
// The external process will send progress updates to the main process through its STDOUT, and the main
// process will forward them to the caller.
sequenceDiagram
    participant MP as Main Process
    participant ES as External Scanner
    participant SP as Subprocess (navidrome scan --subprocess)
    participant FS as File System
    participant DB as Database
    
    Note over MP: DevExternalScanner=true
    MP->>ES: ScanAll(ctx, fullScan)
    activate ES
    
    ES->>ES: Locate executable path
    ES->>SP: Start subprocess with args:<br>scan --subprocess --configfile ... etc.
    activate SP
    
    Note over ES,SP: Create pipe for communication
    
    par Subprocess executes scan
        SP->>FS: Read files & metadata
        SP->>DB: Update database
    and Main process monitors progress
        loop For each progress update
            SP->>ES: Send encoded progress info via stdout pipe
            ES->>MP: Forward progress info
        end
    end
    
    SP-->>ES: Subprocess completes (success/error)
    deactivate SP
    ES-->>MP: Return aggregated warnings/errors
    deactivate ES

Technical details:

  1. Process Isolation

    • Spawns a separate process using the same executable
    • Uses the --subprocess flag to indicate it's running as a child process
    • Preserves configuration by passing required flags (--configfile, --datafolder, etc.)
  2. Inter-Process Communication

    • Uses a pipe for bidirectional communication
    • Encodes/decodes progress updates using Go's gob encoding for efficient binary transfer
    • Properly handles process termination and error propagation
  3. Memory Management Benefits

    • Scanning operations can be memory-intensive, especially with large music libraries
    • Memory leaks or excessive allocations are automatically cleaned up when the process terminates
    • Main Navidrome process remains stable even if scanner encounters memory-related issues
  4. Error Handling

    • Detects non-zero exit codes from the subprocess
    • Propagates error messages back to the main process
    • Ensures resources are properly cleaned up, even in error conditions

Scanning Process Flow

Phase 1: Folder Scan (phase_1_folders.go)

This phase handles the initial traversal and media file processing.

flowchart TD
    A[Start Phase 1] --> B{Full Scan?}
    B -- Yes --> C[Scan All Folders]
    B -- No --> D[Scan Modified Folders]
    C --> E[Read File Metadata]
    D --> E
    E --> F[Create Artists]
    E --> G[Create Albums]
    F --> H[Save to Database]
    G --> H
    H --> I[Mark Missing Folders]
    I --> J[End Phase 1]

Technical implementation details:

  1. Folder Traversal

    • Uses walkDirTree to traverse the directory structure
    • Handles symbolic links and hidden files
    • Processes .ndignore files for exclusions
    • Maps files to appropriate types (audio, image, playlist)
  2. Metadata Extraction

    • Processes files in batches (defined by filesBatchSize = 200)
    • Extracts metadata using the configured storage backend
    • Converts raw metadata to MediaFile objects
    • Collects and normalizes tag information
  3. Album and Artist Creation

    • Groups tracks by album ID
    • Creates album records from track metadata
    • Handles album ID changes by tracking previous IDs
    • Creates artist records from track participants
  4. Database Persistence

    • Uses transactions for atomic updates
    • Preserves album annotations across ID changes
    • Updates library-artist mappings
    • Marks missing tracks for later processing
    • Pre-caches artwork for performance

Phase 2: Missing Tracks Processing (phase_2_missing_tracks.go)

This phase identifies tracks that have moved or been deleted.

flowchart TD
    A[Start Phase 2] --> B[Load Libraries]
    B --> C[Get Missing and Matching Tracks]
    C --> D[Group by PID]
    D --> E{Match Type?}
    E -- Exact --> F[Update Path]
    E -- Same PID --> G[Update If Only One]
    E -- Equivalent --> H[Update If No Better Match]
    F --> I[End Phase 2]
    G --> I
    H --> I

Technical implementation details:

  1. Track Identification Strategy

    • Uses persistent identifiers (PIDs) to track tracks across scans
    • Loads missing tracks and potential matches from the database
    • Groups tracks by PID to limit comparison scope
  2. Match Analysis

    • Applies three levels of matching criteria:
      • Exact match (full metadata equivalence)
      • Single match for a PID
      • Equivalent match (same base path or similar metadata)
    • Prioritizes matches in order of confidence
  3. Database Update Strategy

    • Preserves the original track ID
    • Updates the path to the new location
    • Deletes the duplicate entry
    • Uses transactions to ensure atomicity

Phase 3: Album Refresh (phase_3_refresh_albums.go)

This phase updates album information based on the latest track metadata.

flowchart TD
    A[Start Phase 3] --> B[Load Touched Albums]
    B --> C[Filter Unmodified]
    C --> D{Changes Detected?}
    D -- Yes --> E[Refresh Album Data]
    D -- No --> F[Skip]
    E --> G[Update Database]
    F --> H[End Phase 3]
    G --> H
    H --> I[Refresh Statistics]

Technical implementation details:

  1. Album Selection Logic

    • Loads albums that have been "touched" in previous phases
    • Uses a producer-consumer pattern for efficient processing
    • Retrieves all media files for each album for completeness
  2. Change Detection

    • Rebuilds album metadata from associated tracks
    • Compares album attributes for changes
    • Skips albums with no media files
    • Avoids unnecessary database updates
  3. Statistics Refreshing

    • Updates album play counts
    • Updates artist play counts
    • Maintains consistency between related entities

Phase 4: Playlist Import (phase_4_playlists.go)

This phase imports and updates playlists from the file system.

flowchart TD
    A[Start Phase 4] --> B{AutoImportPlaylists?}
    B -- No --> C[Skip]
    B -- Yes --> D{Admin User Exists?}
    D -- No --> E[Log Warning & Skip]
    D -- Yes --> F[Load Folders with Playlists]
    F --> G{For Each Folder}
    G --> H[Read Directory]
    H --> I{For Each Playlist}
    I --> J[Import Playlist]
    J --> K[Pre-cache Artwork]
    K --> L[End Phase 4]
    C --> L
    E --> L

Technical implementation details:

  1. Playlist Discovery

    • Loads folders known to contain playlists
    • Focuses on folders that have been touched in previous phases
    • Handles both playlist formats (M3U, NSP)
  2. Import Process

    • Uses the core.Playlists service for import
    • Handles both regular and smart playlists
    • Updates existing playlists when changed
    • Pre-caches playlist cover art
  3. Configuration Awareness

    • Respects the AutoImportPlaylists setting
    • Requires an admin user for playlist import
    • Logs appropriate messages for configuration issues

Final Processing Steps

After the four main phases, several finalization steps occur:

  1. Garbage Collection

    • Removes dangling tracks with no files
    • Cleans up empty albums
    • Removes orphaned artists
    • Deletes orphaned annotations
  2. Statistics Refresh

    • Updates artist song and album counts
    • Refreshes tag usage statistics
    • Updates aggregate metrics
  3. Library Status Update

    • Marks scan as completed
    • Updates last scan timestamp
    • Stores persistent ID configuration
  4. Database Optimization

    • Performs database maintenance
    • Optimizes tables and indexes
    • Reclaims space from deleted records

File System Watching

The watcher system (watcher.go) provides real-time monitoring of file system changes:

flowchart TD
    A[Start Watcher] --> B[For Each Library]
    B --> C[Start Library Watcher]
    C --> D[Monitor File Events]
    D --> E{Change Detected?}
    E -- Yes --> F[Wait for More Changes]
    F --> G{Time Elapsed?}
    G -- Yes --> H[Trigger Scan]
    G -- No --> F
    H --> I[Wait for Scan Completion]
    I --> D

Technical implementation details:

  1. Event Throttling

    • Uses a timer to batch changes
    • Prevents excessive rescanning
    • Configurable wait period
  2. Library-specific Watching

    • Each library has its own watcher goroutine
    • Translates paths to library-relative paths
    • Filters irrelevant changes
  3. Platform Adaptability

    • Uses storage-provided watcher implementation
    • Supports different notification mechanisms per platform
    • Graceful fallback when watching is not supported

Edge Cases and Optimizations

Handling Album ID Changes

The scanner carefully manages album identity across scans:

  • Tracks previous album IDs to handle ID generation changes
  • Preserves annotations when IDs change
  • Maintains creation timestamps for consistent sorting

Detecting Moved Files

A sophisticated algorithm identifies moved files:

  1. Groups missing and new files by their Persistent ID
  2. Applies multiple matching strategies in priority order
  3. Updates paths rather than creating duplicate entries

Resuming Interrupted Scans

If a scan is interrupted:

  • The next scan detects this condition
  • Forces a full scan if the previous one was a full scan
  • Continues from where it left off for incremental scans

Memory Efficiency

Several strategies minimize memory usage:

  • Batched file processing (200 files at a time)
  • External scanner process option
  • Database-side filtering where possible
  • Stream processing with pipelines

Concurrency Control

The scanner implements a sophisticated concurrency model to optimize performance:

  1. Phase-Level Parallelism:

    • Phases 1 and 2 run sequentially due to their dependencies
    • Phases 3 and 4 run in parallel using the chain.RunParallel() function
    • Final steps run sequentially to ensure data consistency
  2. Within-Phase Concurrency:

    • Each phase has configurable concurrency for its stages
    • For example, phase_1_folders.go processes folders concurrently: ppl.NewStage(p.processFolder, ppl.Name("process folder"), ppl.Concurrency(conf.Server.DevScannerThreads))
    • Multiple stages can exist within a phase, each with its own concurrency level
  3. Pipeline Architecture Benefits:

    • Producer-consumer pattern minimizes memory usage
    • Work is streamed through stages rather than accumulated
    • Back-pressure is automatically managed
  4. Thread Safety Mechanisms:

    • Atomic counters for statistics gathering
    • Mutex protection for shared resources
    • Transactional database operations

Configuration Options

The scanner's behavior can be customized through several configuration settings that directly affect its operation:

Core Scanner Options

Setting Description Default
Scanner.Enabled Whether the automatic scanner is enabled true
Scanner.Schedule Cron expression or duration for scheduled scans (e.g., "@daily") "0" (disabled)
Scanner.ScanOnStartup Whether to scan when the server starts true
Scanner.WatcherWait Delay before triggering scan after file changes detected 5s
Scanner.ArtistJoiner String used to join multiple artists in track metadata " • "

Playlist Processing

Setting Description Default
PlaylistsPath Path(s) to search for playlists (supports glob patterns) ""
AutoImportPlaylists Whether to import playlists during scanning true

Performance Options

Setting Description Default
DevExternalScanner Use external process for scanning (reduces memory issues) true
DevScannerThreads Number of concurrent processing threads during scanning 5

Persistent ID Options

Setting Description Default
PID.Track Format for track persistent IDs (critical for tracking moved files) "musicbrainz_trackid|albumid,discnumber,tracknumber,title"
PID.Album Format for album persistent IDs (affects album grouping) "musicbrainz_albumid|albumartistid,album,albumversion,releasedate"

These options can be set in the Navidrome configuration file (e.g., navidrome.toml) or via environment variables with the ND_ prefix (e.g., ND_SCANNER_ENABLED=false). For environment variables, dots in option names are replaced with underscores.

Conclusion

The Navidrome scanner represents a sophisticated system for efficiently managing music libraries. Its phase-based pipeline architecture, careful handling of edge cases, and performance optimizations allow it to handle libraries of significant size while maintaining data integrity and providing a responsive user experience.