From b6b9df06dbe8c17ad319242adaf51bfa5aa73c60 Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Sun, 12 Feb 2023 10:00:18 +0100
Subject: [PATCH 1/7] fix(bin/emqx): drop remsh and escript from ps -ef outputs

---
 bin/emqx | 24 ++++++++++++++++++++++--
 1 file changed, 22 insertions(+), 2 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 0a967a73c..911087416 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -464,9 +464,29 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
         export EMQX_BOOT_CONFIGS
     fi
 else
-    # For non-boot commands, we try to get data_dir and ssl_dist_optfile from 'ps -ef' output
+    # For non-boot commands, we need below runtime facts to connect to the running node:
+    #  1. The running node name.
+    #  2. The Erlang cookie in use by the running node name.
+    #  3. SSL options if the node is using TLS for Erlang distribution
+    #
+    # There are 3 sources of truth to get those runtime information.
+    # Listed in the order of preference:
+    #  1. The boot command (which can be inspected from 'ps -ef' command output)
+    #  2. The generated app.<time>.config file located in the dir pointed by 'node.data_dir'
+    #  3. The bootstrap config 'etc/emqx.conf'
+    #
+    # NOTES:
+    #  * We should avoid getting runtime information with the 3rd approach because 'etc/emqx.conf' might
+    #    be updated after the node is started. e.g. If a user starts the node with name 'emqx@127.0.0.1'
+    #    then update the config in the file to 'node.name = "emqx@local.net"', after this change,
+    #    there would be no way stop the running node 'emqx@127.0.0.1', because 'emqx stop' command
+    #    would try to stop the new node instead.
+    #  * The node name and Erlang cookie can be found in 'ps -ef' output, but they are parsed from generated config instead.
+    #  * The primary grep pattern is $RUNNER_ROOT_DIR because one can start multiple nodes at the same time
+    #  * The grep args like '[e]mqx' but not 'emqx' is to avoid greping the grep command itself
+    #  * The running 'remsh' and 'escript' processes must be excluded
     # shellcheck disable=SC2009
-    PS_LINE="$(ps -ef | grep "\-[r]oot $RUNNER_ROOT_DIR" || true)"
+    PS_LINE="$(ps -ef | grep '[e]mqx' | grep -v -E '(remsh|nodetool)' | grep -oE "\-[r]oot ${RUNNER_ROOT_DIR}.*" || true)"
     if [ "$(echo -e "$PS_LINE" | wc -l)" -eq 1 ]; then
         ## only one emqx node is running
         ## strip 'emqx_data_dir ' and ' --' because the dir in between may contain spaces

From d08eb01d90f9e0f83383c993eafb7eeb74120a61 Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Sun, 12 Feb 2023 11:32:06 +0100
Subject: [PATCH 2/7] refactor: read node name and cookie from ps -ef

instead of parsing the generated vm.args file,
because the file might have been deleted
---
 bin/emqx | 72 ++++++++++++++++++++------------------------------------
 1 file changed, 26 insertions(+), 46 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 911087416..853719123 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -299,6 +299,8 @@ fi
 # Make sure log directory exists
 mkdir -p "$RUNNER_LOG_DIR"
 
+# turn off debug as this is static
+set +x
 COMPATIBILITY_CHECK='
     io:format("BEAM_OK~n", []),
     try
@@ -321,14 +323,15 @@ COMPATIBILITY_CHECK='
     end,
     halt(0).
 '
+[ "$DEBUG" -eq 1 ] && set -x
 
 compatiblity_info() {
   # RELEASE_LIB is used by Elixir
   # set crash-dump bytes to zero to ensure no crash dump is generated when erl crashes
   env ERL_CRASH_DUMP_BYTES=0 "$BINDIR/$PROGNAME" \
     -noshell \
-    -boot_var RELEASE_LIB "$ERTS_LIB_DIR/lib" \
     -boot "$REL_DIR/start_clean" \
+    -boot_var RELEASE_LIB "$ERTS_LIB_DIR/lib" \
     -eval "$COMPATIBILITY_CHECK"
 }
 
@@ -464,6 +467,8 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
         export EMQX_BOOT_CONFIGS
     fi
 else
+    # Turn off debug as the ps output can be quite noisy
+    set +x
     # For non-boot commands, we need below runtime facts to connect to the running node:
     #  1. The running node name.
     #  2. The Erlang cookie in use by the running node name.
@@ -481,31 +486,33 @@ else
     #    then update the config in the file to 'node.name = "emqx@local.net"', after this change,
     #    there would be no way stop the running node 'emqx@127.0.0.1', because 'emqx stop' command
     #    would try to stop the new node instead.
-    #  * The node name and Erlang cookie can be found in 'ps -ef' output, but they are parsed from generated config instead.
     #  * The primary grep pattern is $RUNNER_ROOT_DIR because one can start multiple nodes at the same time
     #  * The grep args like '[e]mqx' but not 'emqx' is to avoid greping the grep command itself
-    #  * The running 'remsh' and 'escript' processes must be excluded
+    #  * The running 'remsh' and 'nodetool' processes must be excluded
     # shellcheck disable=SC2009
     PS_LINE="$(ps -ef | grep '[e]mqx' | grep -v -E '(remsh|nodetool)' | grep -oE "\-[r]oot ${RUNNER_ROOT_DIR}.*" || true)"
+    [ "$DEBUG" -eq 1 ] && echo "EMQX processes: $PS_LINE"
     if [ "$(echo -e "$PS_LINE" | wc -l)" -eq 1 ]; then
-        ## only one emqx node is running
-        ## strip 'emqx_data_dir ' and ' --' because the dir in between may contain spaces
-        DATA_DIR="$(echo -e "$PS_LINE" | grep -oE "\-emqx_data_dir.*" | sed -E 's#.+emqx_data_dir[[:blank:]]##g' | sed -E 's#[[:blank:]]--$##g' || true)"
-        if [ "$DATA_DIR" = '' ]; then
-            ## this should not happen unless -emqx_data_dir is not set
-            die "node_is_not_running!" 1
-        fi
-        # get ssl_dist_optfile option
-        SSL_DIST_OPTFILE="$(echo -e "$PS_LINE" | grep -oE '\-ssl_dist_optfile\s.+\s' | awk '{print $2}' || true)"
-        if [ -z "$SSL_DIST_OPTFILE" ]; then
-            EMQX_BOOT_CONFIGS="node.data_dir=${DATA_DIR}\ncluster.proto_dist=inet_tcp"
+        ## only one emqx node is running, get running args from 'ps -ef' output
+        tmp_nodename=$(echo -e "$PS_LINE" | grep -oE "\s\-s?name.*" | awk '{print $2}' || true)
+        tmp_cookie=$(echo -e "$PS_LINE" | grep -oE "\s\-setcookie.*" | awk '{print $2}' || true)
+        tmp_dist="$(echo -e "$PS_LINE" | grep -oE '\-ssl_dist_optfile\s.+\s' | awk '{print $2}' || true)"
+        # data_dir is actually not needed, but kept anyway
+        tmp_daadir="$(echo -e "$PS_LINE" | grep -oE "\-emqx_data_dir.*" | sed -E 's#.+emqx_data_dir[[:blank:]]##g' | sed -E 's#[[:blank:]]--$##g' || true)"
+        if [ -z "$tmp_dist" ]; then
+            tmp_proto='inet_tcp'
         else
-            EMQX_BOOT_CONFIGS="node.data_dir=${DATA_DIR}\ncluster.proto_dist=inet_tls"
+            tmp_proto='inet_tls'
         fi
+        ## Make the format like what call_hocon multi_get prints out, but only need 4 args
+        EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.data_dir=${tmp_daadir}"
+        [ "$DEBUG" -eq 1 ] && echo "EMQX boot-configs: $EMQX_BOOT_CONFIGS"
     else
         ## None or more than one node is running, resolve from boot config
+        ## we have no choiece but to read the bootstrap config (with environment overrides available in the current shell)
         EMQX_BOOT_CONFIGS="$(call_hocon -s "$SCHEMA_MOD" -c "$EMQX_ETC_DIR"/emqx.conf multi_get "${CONF_KEYS[@]}")"
     fi
+    [ "$DEBUG" -eq 1 ] && set -x
 fi
 
 get_boot_config() {
@@ -812,23 +819,6 @@ wait_until_return_val() {
     done
 }
 
-latest_vm_args() {
-    local hint_var_name="$1"
-    local vm_args_file
-    vm_args_file="$(find "$CONFIGS_DIR" -type f -name "vm.*.args" | sort | tail -1)"
-    if [ -f "$vm_args_file" ]; then
-        echo "$vm_args_file"
-    else
-        set +x
-        logerr "Node not initialized?"
-        logerr "Generated config file vm.*.args is not found for command '$COMMAND'"
-        logerr "in config dir: $CONFIGS_DIR"
-        logerr "In case the file has been deleted while the node is running,"
-        logerr "set environment variable '$hint_var_name' to continue"
-        exit 1
-    fi
-}
-
 # backward compatible with 4.x
 tr_log_to_env() {
     local log_to=${EMQX_LOG__TO:-undefined}
@@ -871,6 +861,7 @@ maybe_log_to_console() {
     fi
 }
 
+## To be backward compatible, read and then unset EMQX_NODE_NAME
 if [ -n "${EMQX_NODE_NAME:-}" ]; then
     export EMQX_NODE__NAME="${EMQX_NODE_NAME}"
     unset EMQX_NODE_NAME
@@ -882,13 +873,7 @@ fi
 ## or long name (with '@') e.g. 'emqx@example.net' or 'emqx@127.0.0.1'
 NAME="${EMQX_NODE__NAME:-}"
 if [ -z "$NAME" ]; then
-    if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
-        # for boot commands, inspect emqx.conf for node name
-        NAME="$(get_boot_config 'node.name')"
-    else
-        vm_args_file="$(latest_vm_args 'EMQX_NODE__NAME')"
-        NAME="$(grep -E '^-s?name' "${vm_args_file}" | awk '{print $2}')"
-    fi
+    NAME="$(get_boot_config 'node.name')"
 fi
 
 # force to use 'emqx' short name
@@ -914,18 +899,13 @@ PIPE_DIR="${PIPE_DIR:-/$DATA_DIR/${WHOAMI}_erl_pipes/$NAME/}"
 
 ## Resolve Erlang cookie.
 if [ -n "${EMQX_NODE_COOKIE:-}" ]; then
-    ## To be backward compatible, read EMQX_NODE_COOKIE
+    ## To be backward compatible, read and unset EMQX_NODE_COOKIE
     export EMQX_NODE__COOKIE="${EMQX_NODE_COOKIE}"
     unset EMQX_NODE_COOKIE
 fi
 COOKIE="${EMQX_NODE__COOKIE:-}"
 if [ -z "$COOKIE" ]; then
-    if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
-        COOKIE="$(get_boot_config 'node.cookie')"
-    else
-        vm_args_file="$(latest_vm_args 'EMQX_NODE__COOKIE')"
-        COOKIE="$(grep -E '^-setcookie' "${vm_args_file}" | awk '{print $2}')"
-    fi
+    COOKIE="$(get_boot_config 'node.cookie')"
 fi
 [ -z "$COOKIE" ] && COOKIE="$EMQX_DEFAULT_ERLANG_COOKIE"
 if [ $IS_BOOT_COMMAND = 'yes' ] && [ "$COOKIE" = "$EMQX_DEFAULT_ERLANG_COOKIE" ]; then

From 329a1e2eb18814183fe6b4b774089ef168dfc27c Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Mon, 13 Feb 2023 11:05:55 +0100
Subject: [PATCH 3/7] docs: add change logs

---
 changes/v5.0.18/fix-9961.en.md | 1 +
 changes/v5.0.18/fix-9961.zh.md | 1 +
 2 files changed, 2 insertions(+)
 create mode 100644 changes/v5.0.18/fix-9961.en.md
 create mode 100644 changes/v5.0.18/fix-9961.zh.md

diff --git a/changes/v5.0.18/fix-9961.en.md b/changes/v5.0.18/fix-9961.en.md
new file mode 100644
index 000000000..3d671ed2e
--- /dev/null
+++ b/changes/v5.0.18/fix-9961.en.md
@@ -0,0 +1 @@
+Avoid parsing config files for node name and cookie when executing non-boot commands in bin/emqx
diff --git a/changes/v5.0.18/fix-9961.zh.md b/changes/v5.0.18/fix-9961.zh.md
new file mode 100644
index 000000000..9e75cd6d4
--- /dev/null
+++ b/changes/v5.0.18/fix-9961.zh.md
@@ -0,0 +1 @@
+在 bin/emqx 脚本中，避免在运行非启动命令时解析 emqx.conf 来获取节点名称和 cookie

From cc9ee34eff2a8407ebaafde7b434c14b427069fb Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Mon, 13 Feb 2023 12:17:25 +0100
Subject: [PATCH 4/7] refactor: grep with --color=never

---
 bin/emqx                       | 25 +++++++++++++------------
 changes/v5.0.18/fix-9961.en.md |  2 +-
 changes/v5.0.18/fix-9961.zh.md |  2 +-
 3 files changed, 15 insertions(+), 14 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 853719123..4473efd58 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -230,6 +230,7 @@ usage() {
 }
 
 COMMAND="${1:-}"
+GREP='grep --color=never'
 
 if [ -z "$COMMAND" ]; then
     usage 'help'
@@ -340,7 +341,7 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
     # Read BUILD_INFO early as the next commands may mess up the shell
     BUILD_INFO="$(cat "${REL_DIR}/BUILD_INFO")"
     COMPATIBILITY_INFO="$(compatiblity_info 2>/dev/null || true)"
-    if ! (echo -e "$COMPATIBILITY_INFO" | grep -q 'CRYPTO_OK'); then
+    if ! (echo -e "$COMPATIBILITY_INFO" | $GREP -q 'CRYPTO_OK'); then
         ## failed to start, might be due to missing libs, try to be portable
         export LD_LIBRARY_PATH="${LD_LIBRARY_PATH:-$DYNLIBS_DIR}"
         if [ "$LD_LIBRARY_PATH" != "$DYNLIBS_DIR" ]; then
@@ -349,7 +350,7 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
         ## Turn off debug, because COMPATIBILITY_INFO needs to capture stderr
         set +x
         COMPATIBILITY_INFO="$(compatiblity_info 2>&1 || true)"
-        if ! (echo -e "$COMPATIBILITY_INFO" | grep -q 'BEAM_OK'); then
+        if ! (echo -e "$COMPATIBILITY_INFO" | $GREP -q 'BEAM_OK'); then
             ## not able to start beam.smp
             set +x
             logerr "$COMPATIBILITY_INFO"
@@ -358,7 +359,7 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
             logerr "Version=$REL_VSN"
             logerr "Required dependencies: openssl-1.1.1 (libcrypto), libncurses and libatomic1"
             exit 1
-        elif ! (echo -e "$COMPATIBILITY_INFO" | grep -q 'CRYPTO_OK'); then
+        elif ! (echo -e "$COMPATIBILITY_INFO" | $GREP -q 'CRYPTO_OK'); then
             ## not able to start crypto app
             set +x
             logerr "$COMPATIBILITY_INFO"
@@ -490,22 +491,22 @@ else
     #  * The grep args like '[e]mqx' but not 'emqx' is to avoid greping the grep command itself
     #  * The running 'remsh' and 'nodetool' processes must be excluded
     # shellcheck disable=SC2009
-    PS_LINE="$(ps -ef | grep '[e]mqx' | grep -v -E '(remsh|nodetool)' | grep -oE "\-[r]oot ${RUNNER_ROOT_DIR}.*" || true)"
+    PS_LINE="$(ps -ef | $GREP '[e]mqx' | $GREP -v -E '(remsh|nodetool)' | $GREP -oE "\-[r]oot ${RUNNER_ROOT_DIR}.*" || true)"
     [ "$DEBUG" -eq 1 ] && echo "EMQX processes: $PS_LINE"
     if [ "$(echo -e "$PS_LINE" | wc -l)" -eq 1 ]; then
         ## only one emqx node is running, get running args from 'ps -ef' output
-        tmp_nodename=$(echo -e "$PS_LINE" | grep -oE "\s\-s?name.*" | awk '{print $2}' || true)
-        tmp_cookie=$(echo -e "$PS_LINE" | grep -oE "\s\-setcookie.*" | awk '{print $2}' || true)
-        tmp_dist="$(echo -e "$PS_LINE" | grep -oE '\-ssl_dist_optfile\s.+\s' | awk '{print $2}' || true)"
+        tmp_nodename=$(echo -e "$PS_LINE" | $GREP -oE "\s\-s?name.*" | awk '{print $2}' || true)
+        tmp_cookie=$(echo -e "$PS_LINE" | $GREP -oE "\s\-setcookie.*" | awk '{print $2}' || true)
+        tmp_dist="$(echo -e "$PS_LINE" | $GREP -oE '\-ssl_dist_optfile\s.+\s' | awk '{print $2}' || true)"
         # data_dir is actually not needed, but kept anyway
-        tmp_daadir="$(echo -e "$PS_LINE" | grep -oE "\-emqx_data_dir.*" | sed -E 's#.+emqx_data_dir[[:blank:]]##g' | sed -E 's#[[:blank:]]--$##g' || true)"
+        tmp_datadir="$(echo -e "$PS_LINE" | $GREP -oE "\-emqx_data_dir.*" | sed -E 's#.+emqx_data_dir[[:blank:]]##g' | sed -E 's#[[:blank:]]--$##g' || true)"
         if [ -z "$tmp_dist" ]; then
             tmp_proto='inet_tcp'
         else
             tmp_proto='inet_tls'
         fi
         ## Make the format like what call_hocon multi_get prints out, but only need 4 args
-        EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.data_dir=${tmp_daadir}"
+        EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.data_dir=${tmp_datadir}"
         [ "$DEBUG" -eq 1 ] && echo "EMQX boot-configs: $EMQX_BOOT_CONFIGS"
     else
         ## None or more than one node is running, resolve from boot config
@@ -517,7 +518,7 @@ fi
 
 get_boot_config() {
     path_to_value="$1"
-    echo -e "$EMQX_BOOT_CONFIGS" | grep "$path_to_value=" | sed -e "s/$path_to_value=//g" | tr -d \"
+    echo -e "$EMQX_BOOT_CONFIGS" | $GREP "$path_to_value=" | sed -e "s/$path_to_value=//g" | tr -d \"
 }
 
 EPMD_ARGS="-start_epmd false -epmd_module ekka_epmd -proto_dist ekka"
@@ -744,7 +745,7 @@ generate_config() {
         ARG_KEY=$(echo "$ARG_LINE" | awk '{$NF="";print}')
         ARG_VALUE=$(echo "$ARG_LINE" | awk '{print $NF}')
         ## use the key to look up in vm.args file for the value
-        TMP_ARG_VALUE=$(grep "^$ARG_KEY" "$TMP_ARG_FILE" || true | awk '{print $NF}')
+        TMP_ARG_VALUE=$($GREP "^$ARG_KEY" "$TMP_ARG_FILE" || true | awk '{print $NF}')
         ## compare generated (to override) value to original (to be overridden) value
         if [ "$ARG_VALUE" != "$TMP_ARG_VALUE" ] ; then
             ## if they are different
@@ -769,7 +770,7 @@ is_down() {
     if ps -p "$PID" >/dev/null; then
         # still around
         # shellcheck disable=SC2009 # this grep pattern is not a part of the progra names
-        if ps -p "$PID" | grep -q 'defunct'; then
+        if ps -p "$PID" | $GREP -q 'defunct'; then
             # zombie state, print parent pid
             parent="$(ps -o ppid= -p "$PID" | tr -d ' ')"
             logwarn "$PID is marked <defunct>, parent: $(ps -p "$parent")"
diff --git a/changes/v5.0.18/fix-9961.en.md b/changes/v5.0.18/fix-9961.en.md
index 3d671ed2e..6185a64ea 100644
--- a/changes/v5.0.18/fix-9961.en.md
+++ b/changes/v5.0.18/fix-9961.en.md
@@ -1 +1 @@
-Avoid parsing config files for node name and cookie when executing non-boot commands in bin/emqx
+Avoid parsing config files for node name and cookie when executing non-boot commands in bin/emqx.
diff --git a/changes/v5.0.18/fix-9961.zh.md b/changes/v5.0.18/fix-9961.zh.md
index 9e75cd6d4..edd90b2ca 100644
--- a/changes/v5.0.18/fix-9961.zh.md
+++ b/changes/v5.0.18/fix-9961.zh.md
@@ -1 +1 @@
-在 bin/emqx 脚本中，避免在运行非启动命令时解析 emqx.conf 来获取节点名称和 cookie
+在 bin/emqx 脚本中，避免在运行非启动命令时解析 emqx.conf 来获取节点名称和 cookie。

From be867b13038eb22f3264a986e0d2e3a3497b5355 Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Mon, 13 Feb 2023 12:47:42 +0100
Subject: [PATCH 5/7] refactor: read Erlang kernel app config 'net_ticktime'
 from ps -ef

---
 bin/emqx | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 4473efd58..674610a4f 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -401,8 +401,6 @@ remsh() {
     # Generate a unique id used to allow multiple remsh to the same node
     # transparently
     id="remsh$(relx_gen_id)-${NAME}"
-    # Get the node's ticktime so that we use the same thing.
-    TICKTIME="$(relx_nodetool rpcterms net_kernel get_net_ticktime)"
 
     # shellcheck disable=SC2086
     # Setup remote shell command to control node
@@ -454,7 +452,7 @@ call_hocon() {
 ## Resolve boot configs in a batch
 ## This is because starting the Erlang beam with all modules loaded
 ## and parsing HOCON config + environment variables is a non-trivial task
-CONF_KEYS=( 'node.data_dir' 'node.name' 'node.cookie' 'node.db_backend' 'cluster.proto_dist' )
+CONF_KEYS=( 'node.data_dir' 'node.name' 'node.cookie' 'node.db_backend' 'cluster.proto_dist' 'node.dist_net_ticktime' )
 if [ "$IS_ENTERPRISE" = 'yes' ]; then
     CONF_KEYS+=( 'license.key' )
 fi
@@ -498,6 +496,7 @@ else
         tmp_nodename=$(echo -e "$PS_LINE" | $GREP -oE "\s\-s?name.*" | awk '{print $2}' || true)
         tmp_cookie=$(echo -e "$PS_LINE" | $GREP -oE "\s\-setcookie.*" | awk '{print $2}' || true)
         tmp_dist="$(echo -e "$PS_LINE" | $GREP -oE '\-ssl_dist_optfile\s.+\s' | awk '{print $2}' || true)"
+        tmp_ticktime="$(echo -e "$PS_LINE" | $GREP -oE '\s\-kernel\snet_ticktime\s.+\s' | awk '{print $3}' || true)"
         # data_dir is actually not needed, but kept anyway
         tmp_datadir="$(echo -e "$PS_LINE" | $GREP -oE "\-emqx_data_dir.*" | sed -E 's#.+emqx_data_dir[[:blank:]]##g' | sed -E 's#[[:blank:]]--$##g' || true)"
         if [ -z "$tmp_dist" ]; then
@@ -506,7 +505,7 @@ else
             tmp_proto='inet_tls'
         fi
         ## Make the format like what call_hocon multi_get prints out, but only need 4 args
-        EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.data_dir=${tmp_datadir}"
+        EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.dist_net_ticktime=$tmp_ticktime\nnode.data_dir=${tmp_datadir}"
         [ "$DEBUG" -eq 1 ] && echo "EMQX boot-configs: $EMQX_BOOT_CONFIGS"
     else
         ## None or more than one node is running, resolve from boot config
@@ -523,6 +522,7 @@ get_boot_config() {
 
 EPMD_ARGS="-start_epmd false -epmd_module ekka_epmd -proto_dist ekka"
 PROTO_DIST="$(get_boot_config 'cluster.proto_dist' || true)"
+TICKTIME="$(get_boot_config 'node.dist_net_ticktime' || echo '120')"
 # this environment variable is required by ekka_dist module
 # because proto_dist is overriden to ekka, and there is a lack of ekka_tls module
 export EKKA_PROTO_DIST_MOD="${PROTO_DIST:-inet_tcp}"

From 936eab7fbe5c49e9f1c1c37ad650865aa6efec7e Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Mon, 13 Feb 2023 14:45:21 +0100
Subject: [PATCH 6/7] chore: refine error messages when node is not running

---
 bin/emqx | 29 +++++++++++++++++------------
 1 file changed, 17 insertions(+), 12 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 674610a4f..308aaf51f 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -85,7 +85,7 @@ die() {
 
 assert_node_alive() {
     if ! relx_nodetool "ping" > /dev/null; then
-        die "node_is_not_running!" 1
+        exit 1
     fi
 }
 
@@ -457,6 +457,8 @@ if [ "$IS_ENTERPRISE" = 'yes' ]; then
     CONF_KEYS+=( 'license.key' )
 fi
 
+# Turn off debug as the ps output can be quite noisy
+set +x
 if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
     [ -f "$EMQX_ETC_DIR"/emqx.conf ] || die "emqx.conf is not found in $EMQX_ETC_DIR" 1
     if [ "${EMQX_BOOT_CONFIGS:-}" = '' ]; then
@@ -466,19 +468,21 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
         export EMQX_BOOT_CONFIGS
     fi
 else
-    # Turn off debug as the ps output can be quite noisy
-    set +x
     # For non-boot commands, we need below runtime facts to connect to the running node:
-    #  1. The running node name.
-    #  2. The Erlang cookie in use by the running node name.
-    #  3. SSL options if the node is using TLS for Erlang distribution
+    #  1. The running node name;
+    #  2. The Erlang cookie in use by the running node name;
+    #  3. SSL options if the node is using TLS for Erlang distribution;
+    #  4. Erlang kernel application's net_ticktime config.
     #
     # There are 3 sources of truth to get those runtime information.
     # Listed in the order of preference:
     #  1. The boot command (which can be inspected from 'ps -ef' command output)
-    #  2. The generated app.<time>.config file located in the dir pointed by 'node.data_dir'
+    #  2. The generated vm.<time>.config file located in the dir pointed by 'node.data_dir'
     #  3. The bootstrap config 'etc/emqx.conf'
     #
+    # If failed to read from source 1, the information is retrieved from source 3
+    # i.e. source 2 is never used.
+    #
     # NOTES:
     #  * We should avoid getting runtime information with the 3rd approach because 'etc/emqx.conf' might
     #    be updated after the node is started. e.g. If a user starts the node with name 'emqx@127.0.0.1'
@@ -491,7 +495,8 @@ else
     # shellcheck disable=SC2009
     PS_LINE="$(ps -ef | $GREP '[e]mqx' | $GREP -v -E '(remsh|nodetool)' | $GREP -oE "\-[r]oot ${RUNNER_ROOT_DIR}.*" || true)"
     [ "$DEBUG" -eq 1 ] && echo "EMQX processes: $PS_LINE"
-    if [ "$(echo -e "$PS_LINE" | wc -l)" -eq 1 ]; then
+    running_nodes_count="$(echo -e "$PS_LINE" | wc -l)"
+    if [ "$running_nodes_count" -eq 1 ]; then
         ## only one emqx node is running, get running args from 'ps -ef' output
         tmp_nodename=$(echo -e "$PS_LINE" | $GREP -oE "\s\-s?name.*" | awk '{print $2}' || true)
         tmp_cookie=$(echo -e "$PS_LINE" | $GREP -oE "\s\-setcookie.*" | awk '{print $2}' || true)
@@ -506,14 +511,14 @@ else
         fi
         ## Make the format like what call_hocon multi_get prints out, but only need 4 args
         EMQX_BOOT_CONFIGS="node.name=${tmp_nodename}\nnode.cookie=${tmp_cookie}\ncluster.proto_dist=${tmp_proto}\nnode.dist_net_ticktime=$tmp_ticktime\nnode.data_dir=${tmp_datadir}"
-        [ "$DEBUG" -eq 1 ] && echo "EMQX boot-configs: $EMQX_BOOT_CONFIGS"
     else
         ## None or more than one node is running, resolve from boot config
         ## we have no choiece but to read the bootstrap config (with environment overrides available in the current shell)
         EMQX_BOOT_CONFIGS="$(call_hocon -s "$SCHEMA_MOD" -c "$EMQX_ETC_DIR"/emqx.conf multi_get "${CONF_KEYS[@]}")"
     fi
-    [ "$DEBUG" -eq 1 ] && set -x
 fi
+[ "$DEBUG" -eq 1 ] && echo "EMQX_BOOT_CONFIGS: $EMQX_BOOT_CONFIGS"
+[ "$DEBUG" -eq 1 ] && set -x
 
 get_boot_config() {
     path_to_value="$1"
@@ -889,7 +894,7 @@ case "$NAME" in
 esac
 SHORT_NAME="$(echo "$NAME" | awk -F'@' '{print $1}')"
 if ! (echo "$SHORT_NAME" | grep -q '^[0-9A-Za-z_\-]\+$'); then
-    echo "Invalid node name, should be of format '^[0-9A-Za-z_-]+$'."
+    logerr "Invalid node name, should be of format '^[0-9A-Za-z_-]+$'."
     exit 1
 fi
 # This also changes the program name from 'beam.smp' to node name
@@ -931,7 +936,7 @@ case "${COMMAND}" in
     start)
         # Make sure a node IS not running
         if relx_nodetool "ping" >/dev/null 2>&1; then
-            die "node_is_already_running!"
+            die "Node $NAME is already running!"
         fi
 
         # this flag passes down to console mode

From e0709f7eae8877bd7346d2ac94280c6c6ca3033b Mon Sep 17 00:00:00 2001
From: "Zaiming (Stone) Shi" <zmstone@gmail.com>
Date: Mon, 13 Feb 2023 15:46:23 +0100
Subject: [PATCH 7/7] fix(bin/emqx): ensure protable dynamic libs loaded before
 call_hocon

---
 bin/emqx | 11 +++++------
 1 file changed, 5 insertions(+), 6 deletions(-)

diff --git a/bin/emqx b/bin/emqx
index 308aaf51f..0f6a18437 100755
--- a/bin/emqx
+++ b/bin/emqx
@@ -337,7 +337,7 @@ compatiblity_info() {
 }
 
 # Collect Erlang/OTP runtime sanity and compatibility in one go
-if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
+maybe_use_portable_dynlibs() {
     # Read BUILD_INFO early as the next commands may mess up the shell
     BUILD_INFO="$(cat "${REL_DIR}/BUILD_INFO")"
     COMPATIBILITY_INFO="$(compatiblity_info 2>/dev/null || true)"
@@ -348,11 +348,9 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
             export LD_LIBRARY_PATH="$DYNLIBS_DIR:$LD_LIBRARY_PATH"
         fi
         ## Turn off debug, because COMPATIBILITY_INFO needs to capture stderr
-        set +x
         COMPATIBILITY_INFO="$(compatiblity_info 2>&1 || true)"
         if ! (echo -e "$COMPATIBILITY_INFO" | $GREP -q 'BEAM_OK'); then
             ## not able to start beam.smp
-            set +x
             logerr "$COMPATIBILITY_INFO"
             logerr "Please ensure it is running on the correct platform:"
             logerr "$BUILD_INFO"
@@ -361,14 +359,12 @@ if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
             exit 1
         elif ! (echo -e "$COMPATIBILITY_INFO" | $GREP -q 'CRYPTO_OK'); then
             ## not able to start crypto app
-            set +x
             logerr "$COMPATIBILITY_INFO"
             exit 2
         fi
         logwarn "Using libs from '${DYNLIBS_DIR}' due to missing from the OS."
     fi
-    [ "$DEBUG" -eq 1 ] && set -x
-fi
+}
 
 # Warn the user if ulimit -n is less than 1024
 ULIMIT_F=$(ulimit -n)
@@ -461,6 +457,7 @@ fi
 set +x
 if [ "$IS_BOOT_COMMAND" = 'yes' ]; then
     [ -f "$EMQX_ETC_DIR"/emqx.conf ] || die "emqx.conf is not found in $EMQX_ETC_DIR" 1
+    maybe_use_portable_dynlibs
     if [ "${EMQX_BOOT_CONFIGS:-}" = '' ]; then
         EMQX_BOOT_CONFIGS="$(call_hocon -s "$SCHEMA_MOD" -c "$EMQX_ETC_DIR"/emqx.conf multi_get "${CONF_KEYS[@]}")"
         ## export here so the 'console' command recursively called from
@@ -514,6 +511,8 @@ else
     else
         ## None or more than one node is running, resolve from boot config
         ## we have no choiece but to read the bootstrap config (with environment overrides available in the current shell)
+        [ -f "$EMQX_ETC_DIR"/emqx.conf ] || die "emqx.conf is not found in $EMQX_ETC_DIR" 1
+        maybe_use_portable_dynlibs
         EMQX_BOOT_CONFIGS="$(call_hocon -s "$SCHEMA_MOD" -c "$EMQX_ETC_DIR"/emqx.conf multi_get "${CONF_KEYS[@]}")"
     fi
 fi