diff --git a/CHANGELOG.md b/CHANGELOG.md
index 8c1f0ec9..5394bfd6 100644
--- a/CHANGELOG.md
+++ b/CHANGELOG.md
@@ -2,6 +2,11 @@
 
 This new version of Phoenix.PubSub provides a simpler, more extensible, and more performant Phoenix.PubSub API. For users of Phoenix.PubSub, the API is the same, although frameworks and other adapters will have to migrate accordingly (which often means less code).
 
+## 2.1.4 (2024-09-27)
+
+### Enhancements
+  - Add `:permdown_on_shutdown` option.
+
 ## 2.1.3 (2023-06-14)
 
 ### Bug fixes
diff --git a/lib/phoenix/tracker.ex b/lib/phoenix/tracker.ex
index 676b5e6d..29525d5d 100644
--- a/lib/phoenix/tracker.ex
+++ b/lib/phoenix/tracker.ex
@@ -62,13 +62,34 @@ defmodule Phoenix.Tracker do
   An optional `handle_info/2` callback may also be invoked to handle
   application specific messages within your tracker.
 
-  ## Special Considerations
+  ## Stability and Performance Considerations
 
   Operations within `handle_diff/2` happen *in the tracker server's context*.
   Therefore, blocking operations should be avoided when possible, and offloaded
   to a supervised task when required. Also, a crash in the `handle_diff/2` will
   crash the tracker server, so operations that may crash the server should be
   offloaded with a `Task.Supervisor` spawned process.
+
+  ## Application Shutdown
+
+  A tracker does not automatically replicate its state across the cluster as it
+  shuts down. This means that your supervision tree shuts down normally - as it
+  does when you call `System.stop()` or the BEAM receives a `SIGTERM` - any
+  presences that the local tracker instance has will continue to be seen as
+  present by other trackers in the cluster until the `:down_period` for the
+  instance has passed.
+
+  If you want a normal shutdown to immediately cause other nodes to see that
+  tracker's presences as leaving, pass `permdown_on_shutdown: true`. On the
+  other hand, if you are using `Phoenix.Presence` for clients which will
+  immediately attempt to connect to a new node, it may be preferable to use
+  `permdown_on_shutdown: false`, allowing the disconnected clients time to
+  reconnect before removing their old presences, to avoid overwhelming clients
+  with notifications that many users left and immediately rejoined.
+
+  If the application crashes or is halted non-gracefully (for instance, with a
+  `SIGKILL` or a `Ctrl+C` in `iex`), other nodes will still have to wait the
+  `:down_period` to notice that the tracker's presences are gone.
   """
   use Supervisor
   require Logger
@@ -267,6 +288,9 @@ defmodule Phoenix.Tracker do
     * `:down_period` - The interval in milliseconds to flag a replica
       as temporarily down. Default `broadcast_period * max_silent_periods * 2`
       (30s down detection). Note: This must be at least 2x the `broadcast_period`.
+    * `permdown_on_shutdown` - boolean; whether to immediately call
+      `graceful_permdown/1` on the tracker during a graceful shutdown. See
+      'Application Shutdown' section. Default `false`.
     * `:permdown_period` - The interval in milliseconds to flag a replica
       as permanently down, and discard its state.
       Note: This must be at least greater than the `down_period`.
@@ -287,6 +311,7 @@ defmodule Phoenix.Tracker do
   @impl true
   def init([tracker, tracker_opts, opts, name]) do
     pool_size = Keyword.get(opts, :pool_size, 1)
+    permdown_on_shutdown = Keyword.get(opts, :permdown_on_shutdown, false)
     ^name = :ets.new(name, [:set, :named_table, read_concurrency: true])
     true = :ets.insert(name, {:pool_size, pool_size})
 
@@ -301,13 +326,24 @@ defmodule Phoenix.Tracker do
         }
       end
 
+    children = if permdown_on_shutdown do
+      shards ++ [
+        %{
+          id: :shutdown_handler,
+          start: {Phoenix.Tracker.ShutdownHandler, :start_link, [[tracker: tracker]]}
+        }
+      ]
+    else
+      shards
+    end
+
     opts = [
       strategy: :one_for_one,
       max_restarts: pool_size * 2,
       max_seconds: 1
     ]
 
-    Supervisor.init(shards, opts)
+    Supervisor.init(children, opts)
   end
 
   defp pool_size(tracker_name) do
diff --git a/lib/phoenix/tracker/shutdown_handler.ex b/lib/phoenix/tracker/shutdown_handler.ex
new file mode 100644
index 00000000..584bb7fd
--- /dev/null
+++ b/lib/phoenix/tracker/shutdown_handler.ex
@@ -0,0 +1,21 @@
+defmodule Phoenix.Tracker.ShutdownHandler do
+  @moduledoc false
+  use GenServer
+
+  def start_link(opts) do
+    GenServer.start_link(__MODULE__, opts, name: __MODULE__)
+  end
+
+  @impl GenServer
+  def init(opts) do
+    tracker = Keyword.fetch!(opts, :tracker)
+    Process.flag(:trap_exit, true)
+    {:ok, %{tracker: tracker}}
+  end
+
+  @impl GenServer
+  def terminate(_reason, state) do
+    Phoenix.Tracker.graceful_permdown(state.tracker)
+    :ok
+  end
+end