Module Persistent_connection.Rpc

type t
type address = Core_kernel.Host_and_port.t

The address of a service to which one can connect. E.g. Host_and_port.t is a reasonable choice when making a TCP connection.

type conn = Rpc.Connection.t

A connection, perhaps embellished with additional information upon connection.

module Event : sig ... end
val create : server_name:string -> ?⁠on_event:(Event.t -> unit Async_kernel.Deferred.t) -> ?⁠retry_delay:(unit -> Core_kernel.Time_ns.Span.t) -> ?⁠random_state:Core_kernel.Random.State.t -> ?⁠time_source:Async_kernel.Time_source.t -> connect:(address -> conn Core_kernel.Or_error.t Async_kernel.Deferred.t) -> (unit -> address Core_kernel.Or_error.t Async_kernel.Deferred.t) -> t

create ~server_name ~on_event ~retry_delay get_address returns a persistent connection to a server whose host and port are obtained via get_address every time we try to connect. For example, get_address might look up a server's host and port in catalog at a particular path to which multiple redundant copies of a service are publishing their location. If one copy dies, we get the address of the another one when looking up the address afterwards.

All connection events (see the type above) are passed to the on_event callback, if given. When this callback becomes determined, we move on to the next step in our connection attempt (e.g. we won't actually attempt to connect until on_event Attempting_to_connect is finished). Note that on_event Disconnected will only be called once on_event (Connected conn) finishes even if the connection goes down during that callback.

`Failed_to_connect error and `Obtained_address addr events are only reported if they are distinct from the most recent event of the same type that has taken place since the most recent `Attempting_to_connect event.

Connection is retried after Time.Span.randomize ~percent:(Percent.of_mult 0.3) (retry_delay ()). The default for retry_delay is const (sec 10.). Note that what this retry delay actually throttles is the delay between two connection attempts, so when a long-lived connection dies, connection is usually immediately retried, and if that failed, wait for another retry delay and retry.

The random_state and time_source arguments are there to make persistent connection code more deterministically testable. They default to Random.State.default and Time_source.wall_clock (), respectively.

val connected : t -> conn Async_kernel.Deferred.t

connected returns the first available connection from the time it is called. When currently connected, the returned deferred is already determined. If closed has been called, then the returned deferred is never determined.

val current_connection : t -> conn option

The current connection, if any.

close t closes the current connection and stops it from trying to reconnect. After the deferred it returns becomes determined, the last connection has been closed and no others will be attempted.

Note: no close calls are ever generated internally in response to the connection being closed by the other side.

include Persistent_connection_kernel__.Persistent_connection_kernel_intf.Closable with type t := t
type t

a connection type

val close : t -> unit Async_kernel.Deferred.t

close t closes the connection. The returned deferred becomes determined once any resources needed to maintain the connection have been released.

val is_closed : t -> bool

is_closed t returns true if close has ever been called (even if the returned deferred has not yet been fulfilled).

Note that some modules implementing Closable may call close internally upon noticing that the connection was closed by the other side. The interface of such a module ought to say that this is the case.

val close_finished : t -> unit Async_kernel.Deferred.t

close_finished t becomes determined at the same time as the result of the first call to close. close_finished differs from close in that it does not have the side effect of initiating a close.