Type Members

1.  abstract class CheckedSubscriptionErrorRetryPolicy[SE <: SubscriptionError] extends SubscriptionErrorRetryPolicy

   Allows implementors to only specify policy for an error hierarchy they've defined.

   Allows implementors to only specify policy for an error hierarchy they've defined. Avoids adding type parameters to all sequencer client components. TODO(11067): work out if type parameters are really required and if so are they that bad

2.  trait DelaySequencedEvent extends AnyRef
3.  final case class DelayedSequencerClient(domain: DomainId, member: String) extends DelaySequencedEvent with Product with Serializable
4.  final case class Fatal(msg: String) extends SequencerSubscriptionCreationError with Product with Serializable

   When a fatal error occurs on the creation of a sequencer subscription, the com.digitalasset.canton.sequencing.client.ResilientSequencerSubscription will not retry the subscription creation.

   When a fatal error occurs on the creation of a sequencer subscription, the com.digitalasset.canton.sequencing.client.ResilientSequencerSubscription will not retry the subscription creation. Instead, the subscription will fail.

5.  sealed trait GrpcSequencerAuthenticationSupport extends AnyRef
6.  class PeriodicAcknowledgements extends NamedLogging with FlagCloseable with HasFlushFuture

   Periodically pull the latest clean timestamp and if it has changed acknowledge it with the sequencer.

   Periodically pull the latest clean timestamp and if it has changed acknowledge it with the sequencer. This indicates that we have successfully processed all events up to and including this event. We always acknowledge the current clean timestamp on startup if available to indicate to the sequencer that we are running. The periodic interval is based on the host clock not in sequencer time, however any drift is likely insignificant for the purpose of the sequencer acknowledgements (pruning hourly/daily). Errors are logged at error level - periodic failures are likely not problematic however continuous errors could eventually be problematic for the sequencer operator.

7.  final case class RecordingConfig(directory: Path, filename: Option[String] = None) extends Product with Serializable

   Configuration for where to record sequencer sends and events to.

   Configuration for where to record sequencer sends and events to.

   directory

   Root directory for holding all recording files

   filename

   Filename that is initially empty and updated to a name based on the member-id at runtime. Use setFilename to ensure this can only be set once.

8.  sealed trait ReplayAction extends AnyRef
9.  final case class ReplayConfig(recordingConfig: RecordingConfig, action: ReplayAction) extends Product with Serializable

   Configuration for setting up a sequencer client to replay requests or received events.

   Configuration for setting up a sequencer client to replay requests or received events.

   recordingConfig

   The path to where all recorded content is stored

   action

   What type of replay we'll be performing

10.  trait RequestSigner extends AnyRef
11.  class ResilientSequencerSubscriberPekko[E] extends FlagCloseable with NamedLogging

    Attempts to create a resilient SequencerSubscriptionPekko for the SequencerClient by creating underlying subscriptions using the SequencerSubscriptionFactoryPekko and then recreating them if they fail with a reason that is deemed retryable.

    Attempts to create a resilient SequencerSubscriptionPekko for the SequencerClient by creating underlying subscriptions using the SequencerSubscriptionFactoryPekko and then recreating them if they fail with a reason that is deemed retryable. If a subscription is closed or fails with a reason that is not retryable the failure will be passed downstream from this subscription. We determine whether an error is retryable by calling the SubscriptionErrorRetryPolicy of the supplied SequencerSubscriptionFactoryPekko. We also will delay recreating subscriptions by an interval determined by the com.digitalasset.canton.sequencing.client.SubscriptionRetryDelayRule. The recreated subscription starts at the last event received, or at the starting counter that was given initially if no event was received at all.

    The emitted events stutter whenever the subscription is recreated.

12.  class ResilientSequencerSubscription[HandlerError] extends SequencerSubscription[HandlerError] with NamedLogging with CloseableAtomicHealthComponent with FlagCloseableAsync

    Attempts to create a resilient SequencerSubscription for the SequencerClient by creating underlying subscriptions using the com.digitalasset.canton.sequencing.client.transports.SequencerClientTransport and then recreating them if they fail with a reason that is deemed retryable.

    Attempts to create a resilient SequencerSubscription for the SequencerClient by creating underlying subscriptions using the com.digitalasset.canton.sequencing.client.transports.SequencerClientTransport and then recreating them if they fail with a reason that is deemed retryable. If a subscription is closed or fails with a reason that is not retryable the failure will be passed upstream from this subscription. We determine whether an error is retryable by calling the supplied SubscriptionErrorRetryPolicy. We also will delay recreating subscriptions by an interval determined by the com.digitalasset.canton.sequencing.client.SubscriptionRetryDelayRule. As we have to know where to restart a subscription from when it is recreated we use a com.digitalasset.canton.sequencing.handlers.CounterCapture handler wrapper to keep track of the last event that was successfully provided by the provided handler, and use this value to restart new subscriptions from. For this subscription ResilientSequencerSubscription.start must be called for the underlying subscriptions to begin.

13.  sealed trait SendAsyncClientError extends Product with Serializable with PrettyPrinting

    Errors returned from the AsyncSend where we are sure the request has not potentially been accepted by the server so may be retried using a new message id (as a tracked message id for the failed request may remain in the pending send set).

    Errors returned from the AsyncSend where we are sure the request has not potentially been accepted by the server so may be retried using a new message id (as a tracked message id for the failed request may remain in the pending send set). If a technical error is encountered by the sequencer client where there is a chance that the send will be sequenced it should not be returned to the caller through this error.

14.  type SendCallback = (UnlessShutdown[SendResult]) => Unit

    Signature for callbacks provided to the send operation to take advantage of the SendTracker to provide tracking of the eventual send result.

    Signature for callbacks provided to the send operation to take advantage of the SendTracker to provide tracking of the eventual send result. Callback is ephemeral and will be lost if the SequencerClient is recreated or the process exits.

    See also

    SequencerClient.sendAsync

15.  sealed trait SendResult extends Product with Serializable

    Possible outcomes for a send operation can be observed by a SequencerClient

16.  class SendTracker extends NamedLogging with FlagCloseableAsync with AutoCloseable

    When a we make a send request to the sequencer it will not be sequenced until some point in the future and may not be sequenced at all.

    When a we make a send request to the sequencer it will not be sequenced until some point in the future and may not be sequenced at all. To track a request call send with the messageId and max-sequencing-time of the request, the tracker then observes sequenced events and will notify the provided handler whether the send times out. For aggregatable submission requests, the send tracker notifies the handler of successful sequencing of the submission request, not of successful delivery of the envelopes when the com.digitalasset.canton.sequencing.protocol.AggregationRule.threshold has been reached. In fact, there is no notification of whether the threshold was reached before the max sequencing time.

17.  sealed trait SendType extends AnyRef

    What type of message is being sent.

    What type of message is being sent. Used by the domain and surrounding infrastructure for prioritizing send requests.

18.  sealed trait SequencedEventValidationError[+E] extends Product with Serializable with PrettyPrinting
19.  trait SequencedEventValidator extends AutoCloseable

    Validate whether a received event is valid for processing.

20.  trait SequencedEventValidatorFactory extends AnyRef
21.  class SequencedEventValidatorImpl extends SequencedEventValidator with FlagCloseable with HasCloseContext with NamedLogging

    Validate whether a received event is valid for processing.

22.  trait SequencerClient extends SequencerClientSend with FlagCloseable
23.  final case class SequencerClientConfig(eventInboxSize: PositiveInt = PositiveInt.tryCreate(100), startupConnectionRetryDelay: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofSeconds(1), initialConnectionRetryDelay: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofMillis(10), warnDisconnectDelay: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofSeconds(5), maxConnectionRetryDelay: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofSeconds(30), handshakeRetryAttempts: NonNegativeInt = NonNegativeInt.tryCreate(50), handshakeRetryDelay: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofSeconds(5), defaultMaxSequencingTimeOffset: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofMinutes(5), acknowledgementInterval: NonNegativeFiniteDuration = NonNegativeFiniteDuration.ofMinutes(1), keepAliveClient: Option[KeepAliveClientConfig] = Some(KeepAliveClientConfig()), authToken: AuthenticationTokenManagerConfig = AuthenticationTokenManagerConfig(), optimisticSequencedEventValidation: Boolean = true, skipSequencedEventValidation: Boolean = false, overrideMaxRequestSize: Option[NonNegativeInt] = None, maximumInFlightEventBatches: PositiveInt = PositiveInt.tryCreate(20)) extends Product with Serializable

    Client configured options for how to connect to a sequencer

    Client configured options for how to connect to a sequencer

    eventInboxSize

    The size of the inbox queue used to store received events. Must be at least one. Events in the inbox are processed in parallel. A larger inbox may result in higher throughput at the price of higher memory consumption, larger database queries, and longer crash recovery.

    startupConnectionRetryDelay

    Initial delay before we attempt to establish an initial connection

    initialConnectionRetryDelay

    Initial delay before a reconnect attempt

    warnDisconnectDelay

    Consider sequencer to be degraded after delay

    maxConnectionRetryDelay

    Maximum delay before a reconnect attempt

    handshakeRetryAttempts

    How many attempts should we make to get a handshake response

    handshakeRetryDelay

    How long to delay between attempts to fetch a handshake response

    defaultMaxSequencingTimeOffset

    if no max-sequencing-time is supplied to send, our current time will be offset by this amount

    acknowledgementInterval

    Controls how frequently the client acknowledges how far it has successfully processed to the sequencer which allows the sequencer to remove this data when pruning.

    authToken

    configuration settings for the authentication token manager

    optimisticSequencedEventValidation

    if true, sequenced event signatures will be validated first optimistically and only strict if the optimistic evaluation failed. this means that for a split second, we might still accept an event signed with a key that has just been revoked.

    skipSequencedEventValidation

    if true, sequenced event validation will be skipped. the default setting is false. this option should only be enabled if a defective validation is blocking processing. therefore, unless you know what you are doing, you shouldn't touch this setting.

    overrideMaxRequestSize

    overrides the maxRequestSize configured in the dynamic domain parameters. If overrideMaxRequestSize, is set, modifying the maxRequestSize won't have any effect.

    maximumInFlightEventBatches

    The maximum number of event batches that the system will process concurrently. Setting the maximumInFlightEventBatches parameter limits the number of event batches that the system will process simultaneously, preventing overload and ensuring that the system can handle the workload effectively. A higher value of maximumInFlightEventBatches can lead to increased throughput, but at the cost of higher memory consumption and longer processing times for each batch. A lower value of maximumInFlightEventBatches may limit throughput, but can result in more stable and predictable system behavior.

24.  trait SequencerClientFactory extends AnyRef
25.  class SequencerClientImpl extends SequencerClient with FlagCloseableAsync with NamedLogging with HasFlushFuture with Spanning with HasCloseContext

    The sequencer client facilitates access to the individual domain sequencer.

    The sequencer client facilitates access to the individual domain sequencer. A client centralizes the message signing operations, as well as the handling and storage of message receipts and delivery proofs, such that this functionality does not have to be duplicated throughout the participant node.

26.  trait SequencerClientSend extends AnyRef
27.  sealed trait SequencerClientSubscriptionError extends Product with Serializable
28.  final case class SequencerClientSubscriptionException(error: SequencerClientSubscriptionError) extends RuntimeException with Product with Serializable

    Annotations

    @SuppressWarnings()

29.  trait SequencerClientTransportFactory extends AnyRef
30.  trait SequencerSubscription[HandlerError] extends FlagCloseableAsync with NamedLogging

    A running subscription to a sequencer.

    A running subscription to a sequencer. Can be closed by the consumer or the producer. Once closed the closeReason value will be fulfilled with the reason the subscription was closed. Implementations are expected to immediately start their subscription unless otherwise stated. If close is called while the handler is running closeReason should not be completed until the handler has completed.

31.  sealed trait SequencerSubscriptionCreationError extends SubscriptionError

    Errors that may occur on the creation of a sequencer subscription

32.  trait SequencerSubscriptionFactory[HandlerError] extends AnyRef
33.  trait SequencerSubscriptionFactoryPekko[E] extends AnyRef
34.  final case class SequencerSubscriptionPekko[+E](source: Source[WithKillSwitch[Either[E, OrdinarySerializedEvent]], (KillSwitch, Future[Done])], health: HealthComponent) extends Product with Serializable

    Wrapper for an Pekko source delivering the stream of sequenced events.

    Wrapper for an Pekko source delivering the stream of sequenced events. The org.apache.pekko.stream.KillSwitch can be used to terminate the stream. The materialized scala.concurrent.Future completes after the internal processing in the source has finished after having been closed through the org.apache.pekko.stream.KillSwitch.

35.  final case class SequencerTransportState(transport: SequencerTransportContainer, subscription: Option[Subscription] = None) extends Product with Serializable
36.  class SequencersTransportState extends NamedLogging with FlagCloseable
37.  sealed trait SubscriptionCloseReason[+E] extends AnyRef

    Why did the sequencer subscription terminate

38.  trait SubscriptionErrorRetryPolicy extends AnyRef

    Policy for what errors are considered retryable.

    Policy for what errors are considered retryable. Each transports.SequencerClientTransport is expected to supply their own policy which can consider error types they have defined.

39.  trait SubscriptionErrorRetryPolicyPekko[-E] extends AnyRef

    Policy for what errors are considered retryable.

    Policy for what errors are considered retryable. Each transports.SequencerClientTransport is expected to supply their own policy which can consider error types they have defined.

40.  trait SubscriptionRetryDelayRule extends AnyRef

    Calculator for how to select the next retry duration and specifies what duration is enough to log a warning.