View Source Postgrex.ReplicationConnection behaviour (Postgrex v0.19.1)

A process that receives and sends PostgreSQL replication messages.

Note: this module is experimental and may be subject to changes in the future.

Logical replication

Let's see how to use this module for connecting to PostgreSQL for logical replication. First of all, you need to configure the wal level in PostgreSQL to logical. Run this inside your PostgreSQL shell/configuration:

ALTER SYSTEM SET wal_level='logical';
ALTER SYSTEM SET max_wal_senders='10';
ALTER SYSTEM SET max_replication_slots='10';

Then you must restart your server. Alternatively, you can set those values when starting "postgres". This is useful, for example, when running it from Docker:

services:
  postgres:
    image: postgres:14
    env:
      ...
    command: ["postgres", "-c", "wal_level=logical"]

For CI, GitHub Actions do not support setting command, so you can update and restart Postgres instead in a step:

- name: "Set PG settings"
  run: |
    docker exec ${{ job.services.postgres.id }} sh -c 'echo "wal_level=logical" >> /var/lib/postgresql/data/postgresql.conf'
    docker restart ${{ job.services.pg.id }}

Then you must create a publication to be replicated. This can be done in any session:

CREATE PUBLICATION postgrex_example FOR ALL TABLES;

You can also filter if you want to publish insert, update, delete or a subset of them:

# Skips updates (keeps inserts, deletes, begins, commits, etc)
create PUBLICATION postgrex_example FOR ALL TABLES WITH (publish = 'insert,delete');

# Skips inserts, updates, and deletes (keeps begins, commits, etc)
create PUBLICATION postgrex_example FOR ALL TABLES WITH (publish = '');

Now we are ready to create module that starts a replication slot and listens to our publication. Our example will use the pgoutput for logical replication and print all incoming messages to the terminal:

Mix.install([:postgrex])

defmodule Repl do
  use Postgrex.ReplicationConnection

  def start_link(opts) do
    # Automatically reconnect if we lose connection.
    extra_opts = [
      auto_reconnect: true
    ]

    Postgrex.ReplicationConnection.start_link(__MODULE__, :ok, extra_opts ++ opts)
  end

  @impl true
  def init(:ok) do
    {:ok, %{step: :disconnected}}
  end

  @impl true
  def handle_connect(state) do
    query = "CREATE_REPLICATION_SLOT postgrex TEMPORARY LOGICAL pgoutput NOEXPORT_SNAPSHOT"
    {:query, query, %{state | step: :create_slot}}
  end

  @impl true
  def handle_result(results, %{step: :create_slot} = state) when is_list(results) do
    query = "START_REPLICATION SLOT postgrex LOGICAL 0/0 (proto_version '1', publication_names 'postgrex_example')"
    {:stream, query, [], %{state | step: :streaming}}
  end

  @impl true
  # https://www.postgresql.org/docs/14/protocol-replication.html
  def handle_data(<<?w, _wal_start::64, _wal_end::64, _clock::64, rest::binary>>, state) do
    IO.inspect(rest)
    {:noreply, state}
  end

  def handle_data(<<?k, wal_end::64, _clock::64, reply>>, state) do
    messages =
      case reply do
        1 -> [<<?r, wal_end + 1::64, wal_end + 1::64, wal_end + 1::64, current_time()::64, 0>>]
        0 -> []
      end

    {:noreply, messages, state}
  end

  @epoch DateTime.to_unix(~U[2000-01-01 00:00:00Z], :microsecond)
  defp current_time(), do: System.os_time(:microsecond) - @epoch
end

{:ok, pid} =
  Repl.start_link(
    host: "localhost",
    database: "demo_dev",
    username: "postgres",
  )

Process.sleep(:infinity)

use options

use Postgrex.ReplicationConnection accepts a list of options which configures the child specification and therefore how it runs under a supervisor. The generated child_spec/1 can be customized with the following options:

  • :id - the child specification identifier, defaults to the current module
  • :restart - when the child should be restarted, defaults to :permanent
  • :shutdown - how to shut down the child, either immediately or by giving it time to shut down

For example:

use Postgrex.ReplicationConnection, restart: :transient, shutdown: 10_000

See the "Child specification" section in the Supervisor module for more detailed information. The @doc annotation immediately preceding use Postgrex.ReplicationConnection will be attached to the generated child_spec/1 function.

Name registration

A Postgrex.ReplicationConnection is bound to the same name registration rules as a GenServer. Read more about them in the GenServer docs.

Summary

Types

The following options configure streaming

Callbacks

Invoked after connecting.

Callback for :stream outputs.

Invoked after disconnecting.

Callback for :query outputs.

Callback for process initialization.

Functions

Calls the given replication server.

Returns the integer representation of an LSN value, given its string representation.

Returns the string representation of an LSN value, given its integer representation.

Replies to the given call/3.

Starts a replication process with the given callback module.

Types

@type ack() :: iodata()
@type query() :: iodata()
@type reason() :: String.t()
@type server() :: :gen_statem.server_ref()
@type state() :: term()
@type stream_opts() :: [{:max_messages, pos_integer()}]

The following options configure streaming:

  • :max_messages - The maximum number of replications messages that can be accumulated from the wire until they are relayed to handle_data/2. Defaults to 500.

Callbacks

Link to this callback

handle_call(term, from, state)

View Source (optional)
@callback handle_call(term(), :gen_statem.from(), state()) ::
  {:noreply, state()}
  | {:noreply, ack(), state()}
  | {:query, query(), state()}
  | {:stream, query(), stream_opts(), state()}
  | {:disconnect, reason()}

Callback for call/3.

Replies must be sent with reply/2.

If auto_reconnect: false (the default) and there is a disconnection, the process will terminate and the caller will exit even if no reply is sent. However, if auto_reconnect is set to true, a disconnection will keep the process alive, which means that any command that has not yet been replied to should eventually do so. One simple approach is to reply to any pending commands on handle_disconnect/1.

Link to this callback

handle_connect(state)

View Source (optional)
@callback handle_connect(state()) ::
  {:noreply, state()}
  | {:noreply, ack(), state()}
  | {:query, query(), state()}
  | {:stream, query(), stream_opts(), state()}
  | {:disconnect, reason()}

Invoked after connecting.

This may be invoked multiple times if :auto_reconnect is set to true.

Link to this callback

handle_data(arg1, state)

View Source (optional)
@callback handle_data(binary() | :done, state()) ::
  {:noreply, state()}
  | {:noreply, ack(), state()}
  | {:query, query(), state()}
  | {:stream, query(), stream_opts(), state()}
  | {:disconnect, reason()}

Callback for :stream outputs.

If any callback returns {:stream, iodata, opts, state}, then this callback will be eventually called with the result of the query. It receives binary streaming messages.

This can be useful for replication and copy commands. For replication, the format of the messages are described under the START_REPLICATION section in PostgreSQL docs. Replication messages may require explicit acknowledgement, which can be done by returning a list of binaries according to the replication protocol.

Link to this callback

handle_disconnect(state)

View Source (optional)
@callback handle_disconnect(state()) :: {:noreply, state()}

Invoked after disconnecting.

This is only invoked if :auto_reconnect is set to true.

Link to this callback

handle_info(term, state)

View Source (optional)
@callback handle_info(term(), state()) ::
  {:noreply, state()}
  | {:noreply, ack(), state()}
  | {:query, query(), state()}
  | {:stream, query(), stream_opts(), state()}
  | {:disconnect, reason()}

Callback for Kernel.send/2.

Link to this callback

handle_result(arg1, state)

View Source (optional)
@callback handle_result([Postgrex.Result.t()] | Postgrex.Error.t(), state()) ::
  {:noreply, state()}
  | {:noreply, ack(), state()}
  | {:query, query(), state()}
  | {:stream, query(), stream_opts(), state()}
  | {:disconnect, reason()}

Callback for :query outputs.

If any callback returns {:query, iodata, state}, then this callback will be immediately called with the result of the query. Please note that even though replication connections use the simple query protocol, Postgres currently limits them to single command queries. Due to this constraint, this callback will be passed either a list with a single successful query result or an error.

@callback init(term()) :: {:ok, state()}

Callback for process initialization.

This is called once and before the Postgrex connection is established.

Functions

Link to this function

call(server, message, timeout \\ 5000)

View Source

Calls the given replication server.

@spec decode_lsn(String.t()) :: {:ok, integer()} | :error

Returns the integer representation of an LSN value, given its string representation.

It returns :error if the provided string is not a valid LSN.

Log Sequence Numbers

PostgreSQL uses two representations for the Log Sequence Number (LSN):

  • An unsigned 64-bit integer. Used internally by PostgreSQL and sent in the XLogData replication messages.

  • A string of two hexadecimal numbers of up to eight digits each, separated by a slash. e.g. 1/F73E0220. This is the form accepted by start_replication/2.

For more information on Log Sequence Numbers, see PostgreSQL pg_lsn docs and PostgreSQL WAL internals docs.

@spec encode_lsn(integer()) :: {:ok, String.t()} | :error

Returns the string representation of an LSN value, given its integer representation.

It returns :error if the provided integer falls outside the range for a valid unsigned 64-bit integer.

Log Sequence Numbers

PostgreSQL uses two representations for the Log Sequence Number (LSN):

  • An unsigned 64-bit integer. Used internally by PostgreSQL and sent in the XLogData replication messages.

  • A string of two hexadecimal numbers of up to eight digits each, separated by a slash. e.g. 1/F73E0220. This is the form accepted by start_replication/2.

For more information on Log Sequence Numbers, see PostgreSQL pg_lsn docs and PostgreSQL WAL internals docs.

Replies to the given call/3.

Link to this function

start_link(module, arg, opts)

View Source
@spec start_link(module(), term(), Keyword.t()) ::
  {:ok, pid()} | {:error, Postgrex.Error.t() | term()}

Starts a replication process with the given callback module.

Options

The options that this function accepts are the same as those accepted by Postgrex.start_link/1, except for :idle_interval.

It also accepts extra options for connection management, documented below. Also note this function also automatically set :replication to "database" as part of the connection :parameters if none is set yet.

Connection options

  • :sync_connect - controls if the connection should be established on boot or asynchronously right after boot. Defaults to true.

  • :auto_reconnect - automatically attempt to reconnect to the database in event of a disconnection. See the note about async connect and auto-reconnects above. Defaults to false, which means the process terminates.

  • :reconnect_backoff - time (in ms) between reconnection attempts when :auto_reconnect is enabled. Defaults to 500.