replicore.example.yaml

# Events Streaming Platform service configuration.
events:
  # Events Streaming Platform implementation for the RepliCore control plane to use.
  #
  # Available implementations can be enabled and disabled at compile time so the exact
  # list of options may vary but the following implementations are included by default:
  #
  # - sqlite: store events into a locally persisted SQLite database.
  #   NO SUPPORT FOR HIGH AVAILABLE CLUSTERS.
  #   ONLY SUITABLE FOR SMALL CLUSTERS.
  backend: REQUIRED

  # Implementation specific options are provided as additional attributes here.
  # === For SQLite backend ===
  # Path to the SQLite DB file.
  #path: store.sqlite
  #
  # Events retention and history clean up rules.
  #retention:
  #  # Number of days to keep events in the DB for.
  #  age: 30
  #
  #  # Maximum number of expired events to delete in a single history clean loop.
  #  clean_batch: 500
  #
  #  # Minutes to wait between each run of the history clean loop.
  #  clean_delay: 1

# HTTP Server configuration.
http:
  # Sets the maximum number of pending connections.
  backlog: ~

  # Resolves socket address(es) and binds server to created listener(s).
  bind: "localhost:16016"

  # Maximum time in milliseconds allowed for clients to send all request headers.
  #
  # If a client takes longer to transmit all request headers the request is failed.
  #
  # A value of zero disables the timeout.
  client_request_timeout: ~

  # Server preference for how long to keep connections alive when idle.
  #
  # A value of zero disables keep alive and connections will be
  # closed immediately after the response is sent.
  keep_alive: ~

  # Format of server access logs.
  #
  # Rules for the format string are defined at
  # <https://docs.rs/actix-web/latest/actix_web/middleware/struct.Logger.html#format>.
  log_format: ~

  # Maximum number of concurrent connections for each server worker.
  #
  # This option is available for both TLS and non-TLS modes due to the greatly
  # different CPU requirements.
  #
  # Once the limit is reach listening sockets will stop accepting connections
  # until currently open connections are closed.
  max_connections: ~

  # Maximum number of concurrent TLS connections for each server worker.
  #
  # This option is available for both TLS and non-TLS modes due to the greatly
  # different CPU requirements.
  #
  # Once the limit is reach listening sockets will stop accepting connections
  # until currently open connections are closed.
  max_connections_tls: ~

  # Time in seconds workers are given to complete requests in progress when a shutdown
  # signal is received.
  shutdown_timeout: ~

  # Configure the server to run with TLS encryption.
  tls: ~
  #  # Path to a PEM bundle of Certificate Authorities to verify client certificates with.
  #  #
  #  # When this option is set, clients MUST provide a certificate that is valid.
  #  client_ca_bundle: ~
  #
  #  # Enable TLS for the server.
  #  enabled: true
  #
  #  # Maximum time in milliseconds a TLS handshake must complete in.
  #  #
  #  # If the handshake does not complete in time the connection is closed.
  #  handshake_timeout: ~
  #
  #  # Path to the PEM encoded server private certificate file.
  #  #
  #  # REQUIRED when the TLS block is not empty.
  #  server_private_cert: /path/to/cert.pem
  #
  #  # Path to the PEM encoded server private key file.
  #  #
  #  # REQUIRED when the TLS block is not empty.
  #  server_private_key: /path/to/key.pem

  # Number of workers handling HTTP requests.
  #
  # Defaults to the number of CPUs available.
  workers: ~

# Configuration of the tokio runtime for the process.
#
# These options configure the handling of synchronous and asynchronous tasks.
# These are low level code execution patters and you should be familiar with the concept of
# asynchronous programming before making changes.
#
# For an introduction to async in Rust you can refer to
# <https://rust-lang.github.io/async-book/01_getting_started/02_why_async.html>.
runtime:
  # Allowed time, in seconds, for running operations to complete once process shutdown begins.
  shutdown_grace_sec: 120

  # Maximum number of threads processing blocking tasks.
  #
  # Blocking tasks take over a thread until they complete, even during wait times such as IO.
  # To prevent blocking tasks from preventing non-blocking tasks from executing they get
  # a dedicated pool of threads to execute on.
  #
  # This option sets the maximum number of threads that can run blocking tasks.
  # If all threads are busy, new blocking tasks will be queued until threads are available.
  sync_workers: ~

  # Time in second to keep blocking task threads alive waiting for more tasks.
  sync_workers_keep_alive: ~

  # Number of threads processing non-blocking tasks.
  #
  # As tasks keep a thread busy only when they can progress a small number of threads
  # can handle a large number of non-block tasks.
  #
  # This number is best kept small and defaults to the number of CPU cores on the system.
  workers: ~

# Persistent Store service configuration.
store:
  # Persistent Store implementation for the RepliCore control plane to use.
  #
  # Available implementations can be enabled and disabled at compile time so the exact
  # list of options may vary but the following implementations are included by default:
  #
  # - sqlite: store information into a locally persisted SQLite database.
  #   NO SUPPORT FOR HIGH AVAILABLE CLUSTERS.
  #   ONLY SUITABLE FOR SMALL CLUSTERS.
  backend: REQUIRED

  # Implementation specific options are provided as additional attributes here.
  # === For SQLite backend ===
  # Path to the SQLite DB file.
  #path: store.sqlite

# Configuration for background tasks execution and backend service.
tasks:
  # Tasks executor backoff configuration in case of errors interacting with the Message Queue.
  backoff:
    # Maximum time, in seconds, to wait before retrying after errors from the Message Queue.
    max_delay: 30

    # Maximum number of retries before errors from the Message Queue cause process failure.
    max_retries: 10

    # Backoff multiplier every time a subsequent error is returned by the Message Queue.
    multiplier: 2

    # Initial delay, in milliseconds, to wait before the first retry.
    start_delay: 200

  # Maximum number of tasks to execute concurrently.
  concurrent_tasks: 16  # Actual defaults varies based on available parallelism on the system.

  # Filter queues from which tasks should be processed.
  filters:
    # Ignore subscriptions to any task queue listed here.
    ignore: []

    # If not empty, restrict subscriptions to only queues listed here.
    #
    # If the list is empty all queues can be subscribed to.
    process: []

  # Background Tasks service configuration.
  service:
    # Background Tasks implementation for the RepliCore control plane to use.
    #
    # Available implementations can be enabled and disabled at compile time so the exact
    # list of options may vary but the following implementations are included by default:
    #
    # - sqlite: store tasks into a locally persisted SQLite database.
    #   NO SUPPORT FOR HIGH AVAILABLE CLUSTERS.
    #   ONLY SUITABLE FOR SMALL CLUSTERS.
    backend: REQUIRED

    # Implementation specific options are provided as additional attributes here.
    # === For SQLite backend ===
    # Path to the SQLite DB file.
    #path: store.sqlite

# Telemetry configuration for the process.
telemetry:
  # Logging configuration for the process.
  logs:
    # Emit log events asynchronously.
    #
    # Asynchronous logging can improve performance for but can result
    # in some events loss if the process exists abruptly.
    async: true

    # Only emit log event with this level or grater.
    #
    # Valid options are: CRITICAL, ERROR, WARNING, INFO, DEBUG, TRACE
    # or their lower case versions.
    # Depending on how applications are compiled DEBUG and TRACE logs may
    # not be emitted regardless of what this is set to.
    level: info

    # Set log levels that apply to specific modules only.
    #
    # Valid options are the same as the level field.
    levels:
      # For example you can disable request info logs for HTTP requests only.
      "actix_web::middleware::logger": warning

    # Select how logs are emitted by the process.
    #
    # Valid options are:
    # - JSON: Format logs as a stream of JSON encoded lines to standard out.
    # - TERMINAL: Display logs onto a terminal, with optional colour support.
    mode: json

  # OpenTelemetry configuration for the process.
  otel:
    # Enable export of data using the OpenTelemetry protocol.
    enabled: false

    # GRPC endpoint of the OpenTelemetry agent to send data to.
    endpoint: ~

    # Trace sampling configuration.
    sampling:
      # Follow the sampling decision of the parent span, if any exists.
      follow_parent: true

      # The sampling rule for traces without a parent span.
      #
      # Valid options are: ALWAYS, NEVER, RATIO
      # or their lower case versions.
      #
      # For RATIO mode specify the ratio between 0.0 and 1.0 using an object: {ration: 0.6}.
      mode: ALWAYS

    # Timeout in seconds when communicating with the OpenTelemetry agent.
    timeout_sec: ~

  # Prometheus metrics configuration.
  prom_metrics:
    # Additional labels to attach to all metrics.
    labels: {}

    # Enable collection of process-level metrics (linux only).
    process_metrics: true

  # Sentry error reporting configuration.
  sentry:
    # Sentry DSN (Data Source Name) to send events to.
    #
    # If not set, the environment variable SENTRY_DSN is used.
    dsn: ~

    # Enable sentry integration.
    #
    # Once this option is set to true a DSN must also be specified
    # either above or in SENTRY_DSN for the integration to work.
    enabled: false

    # The ratio of generated events that are submitted to Sentry (between 0.0 and 1.0).
    sample_ratio: 1.0

    # Maximum delay in seconds to process shutdown to flush pending events to Sentry.
    shutdown_timeout: 2