From 9de9631d6b63095ade736036b4b94e2d689d9a40 Mon Sep 17 00:00:00 2001
From: Serge Tupchii <serge.tupchii@protonmail.com>
Date: Tue, 5 Sep 2023 19:29:05 +0300
Subject: [PATCH] fix(nodetool): increase graceful stop timeout, handle and
 report `{badrpc, timeout}` error

---
 bin/nodetool               | 9 ++++++++-
 changes/ce/fix-11567.en.md | 3 +++
 2 files changed, 11 insertions(+), 1 deletion(-)
 create mode 100644 changes/ce/fix-11567.en.md

diff --git a/bin/nodetool b/bin/nodetool
index a96f5f9fd..511ef8e28 100755
--- a/bin/nodetool
+++ b/bin/nodetool
@@ -8,6 +8,8 @@
 %% -------------------------------------------------------------------
 -mode(compile).
 
+-define(SHUTDOWN_TIMEOUT_MS, 120_000).
+
 main(Args) ->
     case os:type() of
         {win32, nt} -> ok;
@@ -85,9 +87,14 @@ do(Args) ->
             %% a "pong"
             io:format("pong\n");
         ["stop"] ->
-            case rpc:call(TargetNode, emqx_machine, graceful_shutdown, [], 60000) of
+            case rpc:call(TargetNode, emqx_machine, graceful_shutdown, [], ?SHUTDOWN_TIMEOUT_MS) of
                 ok ->
                     ok;
+                {badrpc, timeout} ->
+                    io:format("EMQX is still shutting down, it failed to stop gracefully "
+                              "within the configured timeout of: ~ps\n",
+                              [erlang:convert_time_unit(?SHUTDOWN_TIMEOUT_MS, millisecond, second)]),
+                    halt(1);
                 {badrpc, nodedown} ->
                     %% nodetool commands are always executed after a ping
                     %% which if the code gets here, it's because the target node
diff --git a/changes/ce/fix-11567.en.md b/changes/ce/fix-11567.en.md
new file mode 100644
index 000000000..026674f69
--- /dev/null
+++ b/changes/ce/fix-11567.en.md
@@ -0,0 +1,3 @@
+Improve EMQX graceful shutdown (`emqx stop` command):
+- increase timeout from 1 to 2 minutes
+- print an error message if EMQX can't stop gracefully within the configured timeout