#!/bin/sh

# =============================================================================

# alldata.sh —— AllData 平台各服务统一启动 / 停止 / 状态控制脚本

# =============================================================================

#

# 用法:

# sh alldata.sh <command> [target]

#

# command:

# start 启动服务

# stop 停止服务

# restart 重启服务 (等价于 stop + start)

# status 查看服务运行状态

# list 列出全部服务 (分组 / 端口 / 运行状态)

# logs 实时跟踪某个服务的日志 (tail -f)

# help 显示本帮助

#

# target: 省略时默认为 all

# <service> 单个服务, 例如 eureka / gateway / data-server

# <group> 整组服务, 例如 16gmaster / 16gslave / 16gdata

# (可简写为 master / slave / data)

# all 全部服务

#

# 示例:

# sh alldata.sh start eureka # 启动 eureka

# sh alldata.sh stop eureka # 停止 eureka

# sh alldata.sh restart gateway # 重启 gateway

# sh alldata.sh status 16gmaster # 查看 master 组运行状态

# sh alldata.sh start all # 按依赖顺序启动全部服务

# sh alldata.sh list # 列出全部服务

# sh alldata.sh logs data-server # 跟踪 data-server 日志

#

# 可用环境变量:

# JAVA_HOME 指定 JDK, 未设置时使用 PATH 中的 java

# CORE_WAIT 启动 config / eureka 之后等待的秒数(默认 10)

# STOP_TIMEOUT 优雅停止超时秒数, 超时后使用 kill -9(默认 20)

#

# =============================================================================

# 加载系统环境, 保证 java 已在 PATH 中 (沿用原有脚本的做法)

[ -f /etc/profile ] && . /etc/profile 2>/dev/null || true

# 脚本所在目录: 发布包解压后的根目录, 或源码仓库中的 install/

BASE_DIR=$(cd "$(dirname "$0")" && pwd)

CORE_WAIT=${CORE_WAIT:-10}

STOP_TIMEOUT=${STOP_TIMEOUT:-20}

# java 可执行文件: 优先使用 JAVA_HOME

JAVA_BIN=java

if [ -n "$JAVA_HOME" ] && [ -x "$JAVA_HOME/bin/java" ]; then

JAVA_BIN="$JAVA_HOME/bin/java"

fi

# 各服务统一使用的 JVM 参数 (与原有各服务脚本保持一致)

GC_OPTS="-XX:PermSize=128M -XX:MaxPermSize=256M -XX:+UseG1GC -XX:MaxGCPauseMillis=20 -XX:InitiatingHeapOccupancyPercent=35 -XX:+ExplicitGCInvokesConcurrent -XX:MaxInlineLevel=15"

# -----------------------------------------------------------------------------

# 服务注册表

# 字段: <相对目录>|<jar 文件>|<端口>|<profile 或 ->|<Xms>|<Xmx>|<日志文件>

# 目录与日志路径均相对于 BASE_DIR

#

# 说明: 这里把原有 19 个服务脚本的参数集中到一处, 作为唯一事实来源。

# 新增服务只需在此登记, 无需再写单独的启动脚本。

# -----------------------------------------------------------------------------

svc_meta() {

case "$1" in

# ----------------------------- 16gmaster -----------------------------

config)

echo "16gmaster/config|config.jar|8611|-|1024m|2048m|16gmaster/config.log" ;;

system)

echo "16gmaster/data-system-service|data-system-service.jar|8810|dev|128m|2048m|16gmaster/data-system-service/data-system-service.log" ;;

dts)

echo "16gmaster/service-data-dts|service-data-dts.jar|9536|dev|128m|2048m|16gmaster/service-data-dts/service-data-dts.log" ;;

server)

echo "16gmaster/data-server|data-server.jar|8000|dev|128m|2048m|16gmaster/data-server/data-server.log" ;;

market)

echo "16gmaster/data-market-service|data-market-service.jar|8822|dev|128m|2048m|16gmaster/data-market-service/data-market-service.log" ;;

metadata)

echo "16gmaster/data-metadata-service|data-metadata-service.jar|8820|dev|128m|2048m|16gmaster/data-metadata-service/data-metadata-service.log" ;;

market-integration)

echo "16gmaster/data-market-service-integration|data-market-service-integration.jar|8824|dev|128m|2048m|16gmaster/data-market-service-integration/data-market-service-integration.log" ;;

compare)

echo "16gmaster/data-compare-service|data-compare-service.jar|8096|dev|128m|2048m|16gmaster/data-compare-service/data-compare-service.log" ;;

# ----------------------------- 16gslave ------------------------------

eureka)

echo "16gslave/eureka|eureka.jar|8610|-|1024m|2048m|16gslave/eureka.log" ;;

market-mapping)

echo "16gslave/data-market-service-mapping|data-market-service-mapping.jar|8823|-|128m|2048m|16gslave/data-market-service-mapping.log" ;;

masterdata)

echo "16gslave/data-masterdata-service|data-masterdata-service.jar|8828|-|128m|2048m|16gslave/data-masterdata-service.log" ;;

metadata-console)

echo "16gslave/data-metadata-service-console|data-metadata-service-console.jar|8821|-|128m|2048m|16gslave/data-metadata-service-console.log" ;;

quality)

echo "16gslave/data-quality-service|data-quality-service.jar|8826|-|128m|2048m|16gslave/data-quality-service.log" ;;

# ----------------------------- 16gdata -------------------------------

gateway)

echo "16gdata/gateway|gateway.jar|9538|-|1024m|2048m|16gdata/gateway.log" ;;

standard)

echo "16gdata/data-standard-service|data-standard-service.jar|8825|-|128m|2048m|16gdata/data-standard-service.log" ;;

visual)

echo "16gdata/data-visual-service|data-visual-service.jar|8827|-|128m|2048m|16gdata/data-visual-service.log" ;;

quartz)

echo "16gdata/quartz-service|quartz-service.jar|8813|-|128m|2048m|16gdata/quartz-service.log" ;;

*)

return 1 ;;

esac

}

# 从注册表中取某个字段 (1..7)

svc_field() {

svc_meta "$1" | cut -d'|' -f"$2"

}

svc_dir() { svc_field "$1" 1; }

svc_jar() { svc_field "$1" 2; }

svc_port() { svc_field "$1" 3; }

svc_profile() { svc_field "$1" 4; }

svc_xms() { svc_field "$1" 5; }

svc_xmx() { svc_field "$1" 6; }

svc_log() { svc_field "$1" 7; }

# 服务所属分组 (路径第一段)

svc_group() { svc_dir "$1" | cut -d'/' -f1; }

is_service() {

case " $ALL_SERVICES " in

*" $1 "*) return 0 ;;

*) return 1 ;;

esac

}

# -----------------------------------------------------------------------------

# 服务清单与启动顺序

# 启动顺序按依赖排列: 配置中心 config → 注册中心 eureka → 网关 gateway → 其余

# -----------------------------------------------------------------------------

MASTER_SERVICES="config system dts server market metadata market-integration compare"

SLAVE_SERVICES="eureka market-mapping masterdata metadata-console quality"

DATA_SERVICES="gateway standard visual quartz"

ALL_SERVICES="config eureka gateway system dts server market metadata market-integration compare market-mapping masterdata metadata-console quality standard visual quartz"

CORE_SERVICES="config eureka"

# -----------------------------------------------------------------------------

# 输出辅助

# -----------------------------------------------------------------------------

info() { echo "[INFO ] $*"; }

warn() { echo "[WARN ] $*"; }

error() { echo "[ERROR] $*" >&2; }

# 把多行 PID 合并成一行展示 (去掉多余空白)

join_pids() {

echo "$1" | tr '\n' ' ' | sed 's/[[:space:]]*$//'

}

# -----------------------------------------------------------------------------

# 进程查找 / 状态判断

# 以 PID 文件为主(本脚本启动的服务), 以 ps 为兜底(兼容由原有脚本启动的服务)。

# 只用 ps 的话, 在受限容器里 ps 不可用时 start/stop/status 会全部失灵。

# -----------------------------------------------------------------------------

pidfile() {

echo "$BASE_DIR/$(svc_dir "$1")/$1.pid"

}

# 判断 pid 是否为本服务的存活进程: 取得到命令行时做二次校验, 避免 PID 复用误判

pid_alive() {

_pid=$1

_jar=$2

kill -0 "$_pid" 2>/dev/null || return 1

_cmd=$(ps -p "$_pid" -o command= 2>/dev/null)

if [ -n "$_cmd" ]; then

case "$_cmd" in

*"$_jar"*) return 0 ;;

*) return 1 ;;

esac

fi

# ps 不可用, 无法校验, 以 PID 文件为准

return 0

}

find_pids() {

_svc=$1

_jar=$(svc_jar "$_svc")

_pf=$(pidfile "$_svc")

if [ -f "$_pf" ]; then

_p=$(tr -d ' \t\r\n' < "$_pf" 2>/dev/null)

case "$_p" in

''|*[!0-9]*) ;; # 内容非法, 当作没有

*)

if pid_alive "$_p" "$_jar"; then

echo "$_p"

return 0

fi ;;

esac

fi

# 兜底: 通过 ps 查找(例如服务是由原有 xxx.sh 脚本启动的, 没有 PID 文件)

ps -ef 2>/dev/null | grep java | grep -F "$_jar" | grep -v grep | awk '{print $2}'

}

is_running() {

[ -n "$(find_pids "$1")" ]

}

# -----------------------------------------------------------------------------

# 启动单个服务

# -----------------------------------------------------------------------------

start_one() {

_svc=$1

_dir="$BASE_DIR/$(svc_dir "$_svc")"

_jar=$(svc_jar "$_svc")

_port=$(svc_port "$_svc")

_log="$BASE_DIR/$(svc_log "$_svc")"

_profile=$(svc_profile "$_svc")

if [ ! -d "$_dir" ]; then

warn "[$_svc] 目录不存在, 已跳过: $(svc_dir "$_svc")"

return 1

fi

if [ ! -f "$_dir/$_jar" ]; then

warn "[$_svc] 未找到 jar, 已跳过: $(svc_dir "$_svc")/$_jar"

return 1

fi

_pids=$(find_pids "$_svc")

if [ -n "$_pids" ]; then

info "[$_svc] 已在运行 (PID: $(join_pids "$_pids")), 跳过启动"

return 0

fi

if [ "$_profile" = "-" ]; then

_pargs=""

else

_pargs="--spring.profiles.active=$_profile"

fi

# 进入服务目录启动, 使相对路径与原有脚本一致; 启动后立即切回

cd "$_dir" || return 1

nohup "$JAVA_BIN" -jar \

-Xms"$(svc_xms "$_svc")" -Xmx"$(svc_xmx "$_svc")" $GC_OPTS \

"./$_jar" --server.port="$_port" $_pargs \

> "$_log" 2>&1 &

_pid=$!

cd "$BASE_DIR" || return 1

# 等 1 秒确认进程没有立刻退出

sleep 1

if kill -0 "$_pid" 2>/dev/null; then

echo "$_pid" > "$(pidfile "$_svc")"

info "[$_svc] 启动成功 (PID: $_pid, 端口: $_port, 日志: $(svc_log "$_svc"))"

return 0

else

rm -f "$(pidfile "$_svc")"

error "[$_svc] 启动失败, 请查看日志: $(svc_log "$_svc")"

return 1

fi

}

# -----------------------------------------------------------------------------

# 停止单个服务

# 先 SIGTERM 优雅退出, 超时后 SIGKILL

# -----------------------------------------------------------------------------

stop_one() {

_svc=$1

_pf=$(pidfile "$_svc")

_pids=$(find_pids "$_svc")

if [ -z "$_pids" ]; then

rm -f "$_pf"

info "[$_svc] 未在运行, 无需停止"

return 0

fi

info "[$_svc] 正在停止 (PID: $(join_pids "$_pids"))"

# shellcheck disable=SC2086

kill $_pids 2>/dev/null

_n=0

while [ "$_n" -lt "$STOP_TIMEOUT" ]; do

[ -z "$(find_pids "$_svc")" ] && break

sleep 1

_n=$((_n + 1))

done

_left=$(find_pids "$_svc")

if [ -n "$_left" ]; then

warn "[$_svc] 优雅退出超时(${STOP_TIMEOUT}s), 强制 kill -9"

# shellcheck disable=SC2086

kill -9 $_left 2>/dev/null

sleep 1

fi

if [ -z "$(find_pids "$_svc")" ]; then

rm -f "$_pf"

info "[$_svc] 已停止"

return 0

else

error "[$_svc] 停止失败, 仍有进程存活"

return 1

fi

}

# -----------------------------------------------------------------------------

# 状态表格

# 注意: printf 的 %-Ns 按"字节"补齐, 而中文占 2 个显示列宽(3 字节),

# 所以表头需要手工补足空格才能和数据行对齐。

# -----------------------------------------------------------------------------

print_table_header() {

printf ' %s %s %s %s\n' \

"$(printf '服务%30s' '')" \

"$(printf '分组%6s' '')" \

"$(printf '端口%3s' '')" \

"状态"

printf ' %s\n' "--------------------------------------------------------------------------"

}

# -----------------------------------------------------------------------------

# 查看单个服务状态 (简洁一行)

# -----------------------------------------------------------------------------

show_one() {

_svc=$1

_pids=$(find_pids "$_svc")

if [ -n "$_pids" ]; then

printf ' %-34s %-10s %-7s %s\n' \

"$_svc" "$(svc_group "$_svc")" "$(svc_port "$_svc")" \

"运行中 (PID: $(join_pids "$_pids"))"

else

printf ' %-34s %-10s %-7s %s\n' \

"$_svc" "$(svc_group "$_svc")" "$(svc_port "$_svc")" "未运行"

fi

}

# -----------------------------------------------------------------------------

# 目标解析: 把 target 展开成服务列表

# -----------------------------------------------------------------------------

resolve_targets() {

case "$1" in

all|ALL|"")

echo "$ALL_SERVICES" ;;

16gmaster|master)

echo "$MASTER_SERVICES" ;;

16gslave|slave)

echo "$SLAVE_SERVICES" ;;

16gdata|data)

echo "$DATA_SERVICES" ;;

*)

if is_service "$1"; then

echo "$1"

else

error "未知的服务或分组: $1"

echo " 可用服务: $ALL_SERVICES" >&2

echo " 可用分组: 16gmaster | 16gslave | 16gdata | all" >&2

return 1

fi ;;

esac

return 0

}

# -----------------------------------------------------------------------------

# 各命令实现

# -----------------------------------------------------------------------------

cmd_start() {

_targets=$1

_tmp=$(resolve_targets "$_targets") || return 1

[ -n "$_tmp" ] || return 1

_ok=0

_fail=0

_count=$(echo "$_tmp" | wc -w | tr -d ' ')

info "准备启动: $_tmp"

for _svc in $_tmp; do

start_one "$_svc" && _ok=$((_ok + 1)) || _fail=$((_fail + 1))

# 启动多个服务时, 在核心服务(config / eureka)之后就绪等待

if [ "$_count" -gt 1 ]; then

case " $CORE_SERVICES " in

*" $_svc "*)

[ "$CORE_WAIT" -gt 0 ] && sleep "$CORE_WAIT" ;;

esac

fi

done

echo

info "启动完成: 成功 $_ok 个, 失败 $_fail 个 (共 $_count 个)"

[ "$_fail" -eq 0 ]

}

cmd_stop() {

_targets=$1

_tmp=$(resolve_targets "$_targets") || return 1

[ -n "$_tmp" ] || return 1

_ok=0

_fail=0

# 停止时按启动顺序的反序, 避免依赖服务先挂

_rev=""

for _svc in $_tmp; do _rev="$_svc $_rev"; done

info "准备停止: $_tmp"

for _svc in $_rev; do

stop_one "$_svc" && _ok=$((_ok + 1)) || _fail=$((_fail + 1))

done

echo

info "停止完成: 成功 $_ok 个, 失败 $_fail 个"

[ "$_fail" -eq 0 ]

}

cmd_restart() {

_targets=$1

_tmp=$(resolve_targets "$_targets") || return 1

[ -n "$_tmp" ] || return 1

info "准备重启: $_tmp"

for _svc in $_tmp; do

stop_one "$_svc"

done

sleep 2

cmd_start "$_targets"

}

cmd_status() {

_targets=$1

_tmp=$(resolve_targets "$_targets") || return 1

[ -n "$_tmp" ] || return 1

print_table_header

for _svc in $_tmp; do

show_one "$_svc"

done

}

cmd_list() {

print_table_header

for _svc in $ALL_SERVICES; do

show_one "$_svc"

done

}

cmd_logs() {

_svc=$1

if [ -z "$_svc" ]; then

error "logs 命令需要指定服务名, 例如: sh alldata.sh logs eureka"

return 1

fi

if ! is_service "$_svc"; then

error "未知的服务: $_svc"

return 1

fi

_log="$BASE_DIR/$(svc_log "$_svc")"

if [ ! -f "$_log" ]; then

error "日志文件不存在: $(svc_log "$_svc")"

return 1

fi

info "跟踪日志: $(svc_log "$_svc") (Ctrl+C 退出)"

tail -n 100 -f "$_log"

}

usage() {

# 直接复用文件头部的注释作为帮助文本, 避免帮助与实现两处维护

sed -n '2,38p' "$0" | sed 's/^# \{0,1\}//'

}

# -----------------------------------------------------------------------------

# 主流程

# -----------------------------------------------------------------------------

COMMAND=$1

TARGET=$2

[ -z "$TARGET" ] && TARGET=all

case "$COMMAND" in

start) cmd_start "$TARGET" ;;

stop) cmd_stop "$TARGET" ;;

restart) cmd_restart "$TARGET" ;;

status) cmd_status "$TARGET" ;;

list) cmd_list ;;

logs) cmd_logs "$2" ;;

help|-h|--help|"")

usage ;;

*)

error "未知命令: $COMMAND"

echo

usage

exit 1 ;;

esac