From 30ae028774f389f21bdef0a7fa03c90e178cfa03 Mon Sep 17 00:00:00 2001
From: Robert Metzger <rmetzger@apache.org>
Date: Fri, 13 Nov 2020 07:50:57 +0100
Subject: [PATCH] [FLINK-17470] Send sigkill to hanging processes in standalone
 scripts

---
 .../src/main/flink-bin/bin/flink-daemon.sh    | 19 +++++++++++++++++--
 1 file changed, 17 insertions(+), 2 deletions(-)

diff --git a/flink-dist/src/main/flink-bin/bin/flink-daemon.sh b/flink-dist/src/main/flink-bin/bin/flink-daemon.sh
index 50e51d33e5035..cbbc0ce3e6795 100644
--- a/flink-dist/src/main/flink-bin/bin/flink-daemon.sh
+++ b/flink-dist/src/main/flink-bin/bin/flink-daemon.sh
@@ -88,6 +88,21 @@ out="${FLINK_LOG_PREFIX}.out"
 
 log_setting=("-Dlog.file=${log}" "-Dlog4j.configuration=file:${FLINK_CONF_DIR}/log4j.properties" "-Dlog4j.configurationFile=file:${FLINK_CONF_DIR}/log4j.properties" "-Dlogback.configurationFile=file:${FLINK_CONF_DIR}/logback.xml")
 
+function guaranteed_kill {
+  to_stop_pid=$1
+  daemon=$2
+
+  # send sigterm for graceful shutdown
+  kill $to_stop_pid
+  # wait 10 seconds for process to stop. By default, Flink kills the JVM 5 seconds after sigterm.
+  timeout 10 tail --pid=$to_stop_pid -f /dev/null
+  if [ "$?" -eq 124 ]; then
+    echo "Daemon $daemon didn't stop within 10 seconds. Killing it."
+    # send sigkill
+    kill -9 $to_stop_pid
+  fi
+}
+
 case $STARTSTOP in
 
     (start)
@@ -142,7 +157,7 @@ case $STARTSTOP in
 
                 if kill -0 $to_stop > /dev/null 2>&1; then
                     echo "Stopping $DAEMON daemon (pid: $to_stop) on host $HOSTNAME."
-                    kill $to_stop
+                    guaranteed_kill $to_stop $DAEMON
                 else
                     echo "No $DAEMON daemon (pid: $to_stop) is running anymore on $HOSTNAME."
                 fi
@@ -159,7 +174,7 @@ case $STARTSTOP in
             while read to_stop; do
                 if kill -0 $to_stop > /dev/null 2>&1; then
                     echo "Stopping $DAEMON daemon (pid: $to_stop) on host $HOSTNAME."
-                    kill $to_stop
+                    guaranteed_kill $to_stop $DAEMON
                 else
                     echo "Skipping $DAEMON daemon (pid: $to_stop), because it is not running anymore on $HOSTNAME."
                 fi