Neo4j Docker容器重启后挂起并退出的问题排查求助

我创建了一个自定义Docker镜像，通过wrapper脚本启动Neo4j并加载初始数据。首次启动容器时偶尔失败，推测是缓存问题或等待Neo4j启动的时间不足。

核心问题：停止容器后重启时，容器会下载插件，但随后挂起，报错./wrapper.sh: line 57: fg: job has terminated，且重启时/logs/debug.log无日志输出，无法定位是否为权限问题。

我的wrapper脚本 #

#!/bin/bash

# THANK YOU! Special shout-out to @marcellodesales on GitHub
# https://github.com/marcellodesales/neo4j-with-cypher-seed-docker/blob/master/wrapper.sh for such a great example script

# Log the info with the same format as NEO4J outputs
log_info() {
  # https://www.howtogeek.com/410442/how-to-display-the-date-and-time-in-the-linux-terminal-and-use-it-in-bash-scripts/
  # printf '%s %s\n' "$(date -u +"%Y-%m-%d %H:%M:%S:%3N%z") INFO  Wrapper: $1"  # Display UTC time
  printf '%s %s\n' "$(date +"%Y-%m-%d %H:%M:%S:%3N%z") INFO  Wrapper: $1"  # Display local time (PST/PDT)
  return
}

# Adapted from https://github.com/neo4j/docker-neo4j/issues/166#issuecomment-486890785
# Alpine is not supported anymore, so this is newer
# Refactoring: Marcello.deSales+github@gmail.com

# turn on bash's job control
# https://stackoverflow.com/questions/11821378/what-does-bashno-job-control-in-this-shell-mean/46829294#46829294
set -m

# Start the primary process and put it in the background
/docker-entrypoint.sh neo4j &

# Wait for Neo4j
log_info "Checking to see if Neo4j has started at http://${DB_HOST}:${DB_PORT}..."
wget --quiet --tries=20 --waitretry=10 -O /dev/null http://${DB_HOST}:${DB_PORT}
log_info "Neo4j has started 🤓"
log_info "Importing data with auth ${NEO4J_AUTH}"

# Import data
log_info  "Loading and importing Cypher file(s)..."
  for cypherFile in /var/lib/neo4j/import/*.data.cypher; do
      [ -f "$cypherFile" ] || break
      log_info "Running cypher ${cypherFile}"
      cat  ${cypherFile} | bin/cypher-shell -u ${NEO4J_USER} -p ${NEO4J_PASSWORD} --fail-fast --format plain
      log_info  "Renaming import file ${cypherFile}"
      mv  ${cypherFile}  ${cypherFile}.applied
  done

log_info  "Finished loading data"

log_info  "Running startup cypher script..."
  for cypherFile in /var/lib/neo4j/import/*.startup.cypher; do
      [ -f "$cypherFile" ] || break
      log_info "Running cypher ${cypherFile}"
      cat  ${cypherFile} | bin/cypher-shell -u ${NEO4J_USER} -p ${NEO4J_PASSWORD} --fail-fast --format plain
  done
log_info  "Finished running startup script"

# now we bring the primary process back into the foreground
# and leave it there
fg %1

我的Dockerfile #

FROM neo4j

ENV NEO4J_USER=neo4j
ENV NEO4J_PASSWORD=s3cr3t
ENV NEO4J_AUTH=${NEO4J_USER}/${NEO4J_PASSWORD}
ENV NEO4JLABS_PLUGINS='["apoc", "graph-data-science"]'
ENV NEO4J_HOME='/var/lib/neo4j'
ENV DB_HOST='localhost'
ENV DB_PORT=7474
ENV NEO4J_dbms_logs_debug_level='DEBUG'
ENV NEO4J_dbms_logs_user_stdout__enabled='true'

EXPOSE 7474 7473 7687

COPY initial-data/ /var/lib/neo4j/import/

COPY ./docker-scripts/wrapper.sh wrapper.sh

ENTRYPOINT ["./wrapper.sh"]

一、紧急修复：跳过重复初始化逻辑 #

重启时初始数据已导入（文件已重命名为.applied），但脚本仍会等待Neo4j启动，若此时Neo4j进程提前退出就会触发fg报错。添加判断跳过已完成的初始化：

修改wrapper脚本，在启动Neo4j后加入以下逻辑：

# 新增：检查是否已完成数据导入
if ls /var/lib/neo4j/import/*.data.cypher.applied 1> /dev/null 2>&1; then
    log_info "Initial data already imported, skipping data loading step"
else
    # Wait for Neo4j
    log_info "Checking to see if Neo4j has started at http://${DB_HOST}:${DB_PORT}..."
    wget --quiet --tries=20 --waitretry=10 -O /dev/null http://${DB_HOST}:${DB_PORT}
    log_info "Neo4j has started 🤓"
    log_info "Importing data with auth ${NEO4J_AUTH}"

    # Import data
    log_info  "Loading and importing Cypher file(s)..."
      for cypherFile in /var/lib/neo4j/import/*.data.cypher; do
          [ -f "$cypherFile" ] || break
          log_info "Running cypher ${cypherFile}"
          cat  ${cypherFile} | bin/cypher-shell -u ${NEO4J_USER} -p ${NEO4J_PASSWORD} --fail-fast --format plain
          log_info  "Renaming import file ${cypherFile}"
          mv  ${cypherFile}  ${cypherFile}.applied
      done

    log_info  "Finished loading data"
fi

# 启动脚本仍需每次执行
log_info  "Running startup cypher script..."
  for cypherFile in /var/lib/neo4j/import/*.startup.cypher; do
      [ -f "$cypherFile" ] || break
      log_info "Running cypher ${cypherFile}"
      cat  ${cypherFile} | bin/cypher-shell -u ${NEO4J_USER} -p ${NEO4J_PASSWORD} --fail-fast --format plain
  done
log_info  "Finished running startup script"

二、定位Neo4j进程提前退出原因 #

1. 捕获后台进程状态 #

在启动Neo4j后记录PID，定期检查进程是否存活：

# Start the primary process and put it in the background
/docker-entrypoint.sh neo4j &
NEO4J_PID=$!

# 新增：检查Neo4j进程是否存活
check_neo4j_alive() {
    if ! kill -0 $NEO4J_PID 2>/dev/null; then
        log_info "Neo4j process (PID $NEO4J_PID) has exited unexpectedly"
        # 打印Neo4j日志以便排查
        cat /var/lib/neo4j/logs/neo4j.log 2>/dev/null || log_info "No neo4j.log found"
        exit 1
    fi
}

# 在wget等待前后都检查进程状态
check_neo4j_alive
wget --quiet --tries=20 --waitretry=10 -O /dev/null http://${DB_HOST}:${DB_PORT}
check_neo4j_alive

2. 修复日志输出问题 #

重启时debug.log无输出，大概率是权限或配置问题：

• 在Dockerfile中添加权限配置：

  RUN chmod +x wrapper.sh
  # 确保日志目录权限正确
  RUN chown -R neo4j:neo4j /var/lib/neo4j/logs
  

• 修改log_info函数，同时输出到文件和stdout：

  log_info() {
      local msg="$1"
      local timestamp=$(date +"%Y-%m-%d %H:%M:%S:%3N%z")
      local log_line="${timestamp} INFO  Wrapper: ${msg}"
      echo "$log_line"
      echo "$log_line" >> /var/lib/neo4j/logs/wrapper.log
      return
  }
  

3. 优化启动等待逻辑 #

wget仅检查HTTP端口，改用cypher-shell测试Bolt端口（真正就绪状态）：

# 替换原wget等待逻辑
log_info "Waiting for Neo4j Bolt port to be ready..."
until bin/cypher-shell -u ${NEO4J_USER} -p ${NEO4J_PASSWORD} -d system "RETURN 1" --format plain >/dev/null 2>&1; do
    log_info "Neo4j not ready yet, waiting 5 seconds..."
    sleep 5
done
log_info "Neo4j has started 🤓"

三、权限问题排查 #

• 检查容器内文件权限：执行docker exec -it <container_id> ls -l /var/lib/neo4j/，确保import、logs目录所属用户为neo4j
• 在Dockerfile中设置文件权限：

  COPY initial-data/ /var/lib/neo4j/import/
  RUN chown -R neo4j:neo4j /var/lib/neo4j/import/
  
  COPY ./docker-scripts/wrapper.sh wrapper.sh
  RUN chmod +x wrapper.sh && chown neo4j:neo4j wrapper.sh
  

• 避免以root用户运行脚本，在wrapper开头添加：

  if [ "$(id -u)" = "0" ]; then
      exec su - neo4j -c "$0 $@"
  fi
  

内容的提问来源于stack exchange，提问作者gdantimi