Server Listener and Runtime Paths

These settings control where SyncTV listens, how other systems reach it, where runtime files are written, how logs are emitted, and how the management CLI connects.

server

HTTP REST and public gRPC use the same server.port.

server.host

Default:

server:
  host: "0.0.0.0"

Common values:

Value	Meaning	Use case
0.0.0.0	Listen on all IPv4 interfaces	Docker, Kubernetes, servers
127.0.0.1	Localhost only	Local dev or reverse-proxy-only access
::	Listen on all IPv6 interfaces	IPv6 environments

server.port

Default:

server:
  port: 8080

HTTP REST API, public gRPC API, and health checks use this port. In Helm, HTTP and gRPC are exposed through separate Services, but both target the same container port.

server.enable_reflection

Default: false.

Enables public gRPC reflection. It is disabled by default; enable it explicitly for local development or controlled internal debugging, and keep it disabled on public production endpoints.

server.grpc_max_message_size_bytes

Default: 16777216, which is 16 MiB.

This limits gRPC request and response message sizes. Increase only when large batch operations require it. Avoid very high values on public endpoints.

server.grpc_compression_enabled

Default:

server:
  grpc_compression_enabled: true

Allows the main SyncTV process gRPC clients and servers to negotiate gzip compression, including public gRPC, management gRPC, cluster/livestream internal gRPC links, and outbound remote media-provider gRPC clients. It does not force every request to be compressed; compression is used when the peer advertises support and the message can benefit from it.

Standalone remote media-provider server deployments do not load the main SyncTV config file. Configure their server-side compression negotiation with PROVIDER_GRPC_COMPRESSION_ENABLED instead.

Keep it enabled when:

• Cluster gRPC, remote management, batch responses, or list APIs are common.
• Network bandwidth is tighter than CPU.
• Kubernetes or Ingress traffic crosses nodes.

Disable it only when CPU is the tighter resource and gRPC payloads are usually small or entirely local.

Environment variable:

SYNCTV_SERVER_GRPC_COMPRESSION_ENABLED=true

server.trusted_proxies

Default: [].

Controls which reverse proxies may provide trusted client IP headers such as X-Forwarded-For.

server:
  trusted_proxies:
    - "10.0.0.0/8"
    - "172.16.0.0/12"
    - "192.168.0.0/16"

Do not use 0.0.0.0/0 or ::/0. If unsure, keep the list empty so SyncTV uses the socket peer address.

server.cors_allowed_origins

Default: [].

Allowed browser origins for cross-origin API access:

server:
  cors_allowed_origins:
    - "https://app.example.com"
    - "https://admin.example.com"

Values must be origins only: scheme, host, and optional port. Do not include paths, query strings, or fragments.

Environment variable form:

SYNCTV_SERVER_CORS_ALLOWED_ORIGINS='["https://app.example.com"]'

server.advertise_host

Address that other nodes should use to reach this node.

Common cases:

• Kubernetes: set through SYNCTV_SERVER_ADVERTISE_HOST or downward API using Pod IP.
• Static multi-node: set an internal IP or DNS name.
• Single-node: usually leave empty.

When empty, SyncTV tries POD_IP, then system hostname.

server.shutdown_drain_timeout_seconds

Default: 30.

How long shutdown waits for existing connections to drain. Increase it for many long-lived connections or smoother Kubernetes rolling updates.

Timezone

time.timezone defaults to automatic detection.

Resolution order:

1. time.timezone
2. SYNCTV_TIME_TIMEZONE
3. TZ
4. System timezone
5. UTC

Example:

time:
  timezone: "Asia/Shanghai"

data_dir

data_dir is the root for runtime-owned local files:

• Management Unix socket.
• File logs.
• HLS livestream segments.
• Proxy slice cache files.

Examples:

data_dir: "/var/lib/synctv"

Container deployments usually use /data and mount it as a volume.

data_dir does not affect *_file secret paths or metrics TLS cert/key paths.

Logging

Top-level logging is the global catch-all output for the database, Redis, caches, proxies, realtime, outbox, migrations, startup, third-party dependencies, and all remaining targets. server, health, metrics, management, cluster, livestream, and webrtc use dedicated outputs. Each event enters exactly one output: dedicated targets enter their component output, and every remaining target enters the global output.

Field	Default	Purpose
logging.level	info	Global log level
logging.format	text	Global log format (text or json)
logging.output	stdout	Global stdout, stderr, or file output
logging.color	auto	Global color policy (auto, always, or never)
<component>.logging.level	info (warn for metrics/cluster)	Component log level
<component>.logging.format	text	Component log format (text or json)
<component>.logging.output	stdout	Independent stdout, stderr, or file output
<component>.logging.color	auto	auto, always, or never

Use json in Kubernetes or log aggregation systems:

logging:
  level: "info"
  format: "json"
  output: stdout
  color: never

server:
  logging:
    level: "info"
    format: "json"
    output: stdout
    color: never

livestream:
  logging:
    level: "info"
    format: "json"
    output:
      type: file
      path: "logs/livestream.log"
      rotation:
        strategy: daily
        max_files: 30
    color: never

webrtc:
  logging:
    level: "info"
    format: "json"
    output: stdout
    color: never

Relative file paths for global and component logs resolve under data_dir. File outputs support daily, hourly, and never rotation strategies; rotation.max_files retains 30 files by default. Containers commonly send every output to stdout for JSON-aware log collection.

Management Endpoint

The management endpoint is used by the synctv CLI for controlled administrative operations. It is not a client-facing business API.

Field	Default	Purpose
management.enabled	true	Enable management endpoint
management.transport	Unix on Unix-like systems, TCP elsewhere	Transport mode
management.port	50052	TCP management port
management.unix_socket_path	platform-specific under data_dir	Unix socket path
management.auth_token	""	Bearer token for TCP management
management.enable_reflection	false	Management gRPC reflection

TCP management requires a strong auth token and should not be exposed publicly.

Unix socket paths can be relative to data_dir. TCP mode is useful for Windows or controlled remote management networks.