Docker启动Spring Boot应用出现Segmentation Fault（错误码139）求助

问题场景 #

我有一个名为users的Spring Boot应用，通过Docker Compose启动时，应用容器直接报Segmentation fault并以错误码139退出，相关日志及配置文件如下：

启动日志 #

Starting docker compose
Container users-user-db-1  Creating
Container users-solace-1  Creating
Container users-solace-1  Created
Container users-user-db-1  Created
Container users-users-1  Creating
Container users-users-1  Created
Attaching to users-users-1, users-solace-1, users-user-db-1
users-user-db-1  | The files belonging to this database system will be owned by user "postgres".
users-user-db-1  | This user must also own the server process.
users-user-db-1  |
users-user-db-1  | The database cluster will be initialized with locale "en_US.utf8".
users-user-db-1  | The default database encoding has accordingly been set to "UTF8".
users-user-db-1  | The default text search configuration will be set to "english".
users-user-db-1  |
users-user-db-1  | Data page checksums are disabled.
users-user-db-1  |
users-user-db-1  | fixing permissions on existing directory /var/lib/postgresql/data ... ok
users-user-db-1  | creating subdirectories ... ok
users-user-db-1  | selecting dynamic shared memory implementation ... posix
users-user-db-1  | selecting default max_connections ... 100
users-solace-1              | Host Boot ID: 
users-solace-1              | Starting VMR Docker Container: Wed Apr 5 08:21:32 UTC 2023
users-solace-1              | Setting umask to 077
users-solace-1              | SolOS Version: soltr_9.9.0.34
users-user-db-1  | selecting default shared_buffers ... 128MB
users-user-db-1  | selecting default time zone ... Etc/UTC
users-user-db-1  | creating configuration files ... ok
users-user-db-1  | running bootstrap script ... ok
users-users-1          | Segmentation fault
users-users-1 exited with code 0

相关配置文件 #

docker-compose.sh

#!/bin/bash

docker container rm -f $(docker container ls -aq)

mvn clean install -DskipTests=true -Dmaven.javadoc.skip=true

docker build --no-cache -f Dockerfile.local -t users:latest .

docker-compose -f docker-compose.yml up --build

Dockerfile.local

FROM <url>/openjdk:11-jdk
MAINTAINER users

ARG JAR_FILE=target/*.jar

COPY ${JAR_FILE} app.jar

CMD java -jar /app.jar

EXPOSE 8080

docker-compose.yml

version: '3'
services:
  users:
    image: users:latest
    restart: on-failure
    networks:
      default:
        aliases:
          - users.svc
    links:
      - user-db
      - solace
    ports:
      - "8094:8080"
    depends_on:
      - user-db
      - solace
    environment:
      WAIT_HOSTS: solace.svc:8080, user-db.svc:5432
      USERS_DB_JDBC_URL: jdbc:postgresql://user-db.svc:5432/users
      USERS_DB_APP_USER_NAME: USERS_rwuser
      USERS_DB_APP_USER_PASSWORD: mysecretpassword
      USERS_DB_MIGRATION_USER_NAME: USERS_rwuser
      USERS_DB_MIGRATION_USER_PASSWORD: mysecretpassword
      USERS_DB_READ_WRITE_USER_ROLE: USERS_rwuser
      USERS_DB_READ_ONLY_USER_ROLE: USERS_rwuser
      SOLACE_HOST_URL: tcp://solace.svc:55555
      SOLACE_MSG_VPN: default
      USERS_SOLACE_USERNAME: admin
      USERS_SOLACE_PASSWORD: admin
    command: >
          java -jar /app.jar"

  user-db:
    image: <URL>/postgres:12.4
    restart: always
    networks:
      default:
        aliases:
          - user-db.svc
    ports:
      - 5432:5432
    environment:
      POSTGRES_USER: USERS_rwuser
      POSTGRES_PASSWORD: mysecretpassword
      POSTGRES_DB: users

  solace:
    image: <URL>/solace/solace-pubsub-standard:9.9.0.34
    restart: always
    networks:
      default:
        aliases:
          - solace.svc
    shm_size: 1g
    ulimits:
      core: 1
      nofile:
        soft: 2448
        hard: 38048
    environment:
      - system_scaling_maxconnectioncount=100

解决方案 #

1. 修复docker-compose.yml中的语法错误 #

查看users服务的command配置，末尾存在多余的双引号，会导致命令解析异常：

command: >
      java -jar /app.jar"

修正方式：去掉多余引号，或者直接删除整个command字段（Dockerfile中已定义CMD，重复定义可能引发冲突）：

# 修正后的command配置
command: >
      java -jar /app.jar

2. 更换可靠的JDK基础镜像 #

当前使用的<url>/openjdk:11-jdk可能存在兼容性问题，建议替换为官方维护的镜像：

# 官方OpenJDK轻量镜像
FROM openjdk:11-jdk-slim
# 或使用Adoptium维护的镜像
FROM eclipse-temurin:11-jdk

3. 确保依赖服务完全就绪 #

depends_on仅保证容器启动，不保证服务内部就绪。当前WAIT_HOSTS配置的Solace端口是8080（管理端口），但应用实际连接的是55555业务端口，建议调整：

WAIT_HOSTS: solace.svc:55555, user-db.svc:5432

也可使用wait-for-it脚本替代原生等待逻辑，确保PostgreSQL和Solace服务完全启动后再启动Spring Boot应用。

4. 调整JVM内存参数 #

Segmentation Fault可能与内存分配异常有关，在启动命令中添加JVM内存限制参数：

command: >
      java -Xmx512m -Xms256m -jar /app.jar

可根据应用实际需求调整-Xmx（最大堆内存）和-Xms（初始堆内存）的值。

5. 启用核心转储排查深层问题 #

在users服务中添加ulimits配置开启核心转储，方便定位崩溃点：

users:
  # 其他配置...
  ulimits:
    core: unlimited

启动后若再次崩溃，可进入容器使用gdb分析核心文件：

# 进入容器
docker exec -it users-users-1 bash
# 分析核心文件
gdb java core.<PID>

内容的提问来源于stack exchange，提问作者user2800089