Linux下MySQL数据库故障系统化诊断的脚本分享

#!/bin/bash
#===============================================================================
# 脚本名称: mysql_crash_diag.sh
# 
#===============================================================================
set -o pipefail
# -----------------------------【单段可复制终端执行】全局常量、颜色、工具函数 -----------------------------
# 入参与时间戳
OUT_DIR="${1:-/tmp/mysql_diag_$(date +%Y%m%d_%H%M%S)}"
TIMESTAMP=$(date '+%Y-%m-%d %H:%M:%S')
mkdir -p "$OUT_DIR"
# 统一存放所有可复制关键路径、PID、故障标记
KEY_COPY_FILE="${OUT_DIR}/key_copy_info.txt"
> "$KEY_COPY_FILE"
echo "===== 全部可复制关键信息汇总（PID/日志/配置/数据目录） =====" > "$KEY_COPY_FILE"
# 颜色定义
RED='\033[0;31m'
GREEN='\033[0;32m'
YELLOW='\033[1;33m'
WHITE='\033[1;37m'
CYAN='\033[0;36m'
NC='\033[0m'
# 日志打印函数
info() { echo -e "${GREEN}[INFO]${NC} $*"; echo "[正常] $*" >> "$KEY_COPY_FILE"; }
warn() { echo -e "${YELLOW}[WARN]${NC} $*"; echo "[警告] $*" >> "$KEY_COPY_FILE"; }
err()  { echo -e "${RED}[ERROR]${NC} $*"; echo "[严重异常] $*" >> "$KEY_COPY_FILE"; }
section() { echo -e "\n${YELLOW}========== $1 ==========${NC}"; echo -e "\n【诊断模块】$1" >> "$KEY_COPY_FILE"; }
# 核心：打印高亮可复制变量，自动写入清单
print_copy() {
    local label="$1"
    local val="$2"
    echo -e "${WHITE}【可复制 $label 】: ${RED}$val${NC}"
    echo "$label = $val" >> "$KEY_COPY_FILE"
}
# 输出可直接执行的完整命令
print_cmd() {
    echo -e "${CYAN}[一键执行命令]${NC} $1"
    echo "执行命令：$1" >> "$KEY_COPY_FILE"
}
# MySQL 全局路径池
MYSQL_LOG_PATHS=( "/var/log/mysqld.log" "/var/log/mysql/error.log" "/var/lib/mysql/*.err" )
MYSQL_CNF_PATHS=( "/etc/my.cnf" "/etc/mysql/my.cnf" "/usr/local/mysql/my.cnf" )
MYSQL_DATA_PATHS=( "/var/lib/mysql" "/usr/local/mysql/data" )
# 全局缓存变量（全程复用，所有命令自动带入）
MYSQL_PID=""
FOUND_ERR_LOG=""
SELECTED_CNF=""
DATA_DIR=""
CRASH_TIMESTAMP=""
IS_DOCKER_MYSQL=0
DOCKER_MYSQL_CID=""
# ----------------------------------------------------------------------------------------------------------
# -----------------------------【单段可复制终端执行】0. 部署模式自动识别：宿主机 / Docker MySQL -----------------------------
section "0. 识别MySQL部署模式（宿主机systemd / Docker容器）"
# 先判断宿主机本地MySQL进程
MYSQL_PID=$(pgrep -x mysqld | head -1 || pgrep -x mariadbd | head -1)
if [[ -n "$MYSQL_PID" ]]; then
    info "检测到宿主机本地MySQL进程，PID已缓存"
    print_copy "MySQL主进程PID" "$MYSQL_PID"
    # 自动解析数据目录
    DATA_DIR=$(ps -ef | grep mysqld | grep -oP '--datadir=\K\S+' | head -1)
    [[ -z "$DATA_DIR" ]] && DATA_DIR="/var/lib/mysql"
    print_copy "MySQL数据目录" "$DATA_DIR"
else
    # 无本地进程，检查Docker内MySQL容器
    DOCKER_MYSQL_CID=$(docker ps -a --filter "name=mysql" --filter "name=mariadb" -q | head -1 2>/dev/null || true)
    if [[ -n "$DOCKER_MYSQL_CID" ]]; then
        IS_DOCKER_MYSQL=1
        warn "未检测到宿主机MySQL，识别到Docker MySQL容器"
        print_copy "Docker MySQL容器CID" "$DOCKER_MYSQL_CID"
        print_cmd "docker logs $DOCKER_MYSQL_CID --tail 300"
        print_cmd "docker exec -it $DOCKER_MYSQL_CID cat \$(mysql -sse \"SELECT @@log_error;\")"
    else
        warn "未找到宿主机MySQL进程、无MySQL容器，请手动确认部署方式"
        print_cmd "ps aux | grep -E mysqld|mariadb"
        print_cmd "docker ps -a | grep -E mysql|mariadb"
    fi
fi
echo "[$TIMESTAMP] 诊断报告生成中，全部数据存放目录: $OUT_DIR" | tee "$OUT_DIR/report.log"
print_copy "诊断文件根目录" "$OUT_DIR"
print_cmd "cat $KEY_COPY_FILE # 查看所有可复制PID、路径汇总"
# ==========================================================================================================
# 步骤1：现象取证与时间锚定（本体论 & 整体论）【单段可复制终端执行】
# ==========================================================================================================
section "1. 现象取证，保护现场快照（内核/服务/进程）"
# 1.1 内核环形缓冲（OOM、硬件IO崩溃底层根源）
dmesg > "$OUT_DIR/dmesg_full.log" 2>/dev/null || echo "无法保存dmesg，手工执行: dmesg > dmesg_manual.log" | tee -a "$OUT_DIR/report.log"
# 单独提取OOM与磁盘错误，单独存档
grep -i "oom-killer\|killed process\|i/o error\|ro filesystem" "$OUT_DIR/dmesg_full.log" > "$OUT_DIR/dmesg_oom_io_err.log"
info "内核OOM/IO错误快照保存: $OUT_DIR/dmesg_oom_io_err.log"
# 1.2 systemd服务状态快照（宿主机模式）
if [[ $IS_DOCKER_MYSQL -eq 0 ]]; then
    if command -v systemctl &>/dev/null; then
        systemctl status mysqld 2>/dev/null > "$OUT_DIR/mysqld_systemd.log" || systemctl status mariadb > "$OUT_DIR/mysqld_systemd.log" 2>/dev/null
        journalctl -u mysqld --since "10 min ago" --no-pager > "$OUT_DIR/mysqld_journal.log" 2>/dev/null
    else
        service mysql status > "$OUT_DIR/mysqld_service.log" 2>/dev/null
    fi
else
    # Docker模式：抓取容器启停日志
    docker logs "$DOCKER_MYSQL_CID" --tail 500 > "$OUT_DIR/docker_mysql_container_log.log" 2>/dev/null
fi
# 1.3 全量进程快照
ps aux > "$OUT_DIR/all_process_snapshot.log"
ps aux | grep -v grep | grep -E mysqld|mariadb > "$OUT_DIR/mysql_process.log"
info "MySQL进程快照已保存"
# 1.4 端口占用检查（常见启动失败原因）
ss -tlnp | grep 3306 > "$OUT_DIR/port_3306_occupy.log"
if [[ -s "$OUT_DIR/port_3306_occupy.log" ]]; then
    warn "检测到3306端口占用，查看文件确认占用进程"
    cat "$OUT_DIR/port_3306_occupy.log"
fi
# 1.5 自动提取崩溃时间戳（因果链锚点）
CRASH_TIMESTAMP=""
# 从systemd日志提取停机时间
if [[ -f "$OUT_DIR/mysqld_systemd.log" ]]; then
    CRASH_TIMESTAMP=$(grep -i "Active: inactive\|stop\|failed" "$OUT_DIR/mysqld_systemd.log" | head -1 | grep -oP '\d{4}-\d{2}-\d{2} \d{2}:\d{2}:\d{2}')
fi
# 内核OOM时间兜底
if [[ -z "$CRASH_TIMESTAMP" && -s "$OUT_DIR/dmesg_oom_io_err.log" ]]; then
    CRASH_TIMESTAMP=$(grep -i "mysql\|mariadb" "$OUT_DIR/dmesg_oom_io_err.log" | tail -1 | grep -oP '\[\s*\d+\.\d+\]' | head -1)
fi
print_copy "推断崩溃关键时间戳" "${CRASH_TIMESTAMP:-未识别，请人工核对日志时间}"
echo "推断的关键异常时间点: ${CRASH_TIMESTAMP:-未自动识别，请手工检查日志时间戳}" | tee -a "$OUT_DIR/report.log"
# 手工检验可复制命令
print_cmd "ls -lt /var/log/mysqld.log /var/log/messages"
print_cmd "journalctl -u mysqld --since \"$CRASH_TIMESTAMP\" --no-pager"
# ==========================================================================================================
# 步骤2：直达本质——错误日志本体分析【单段可复制终端执行】
# ==========================================================================================================
section "2. 扫描MySQL错误日志，提取底层崩溃根源"
# 自动检索真实错误日志路径
for LOG in "${MYSQL_LOG_PATHS[@]}"; do
    for F in $LOG; do
        if [ -f "$F" ]; then
            FOUND_ERR_LOG="$F"
            break 2
        fi
    done
done
# Docker模式特殊读取容器内日志
if [[ $IS_DOCKER_MYSQL -eq 1 && -z "$FOUND_ERR_LOG" ]]; then
    warn "宿主机未找到日志，从Docker容器内抓取错误日志"
    docker exec "$DOCKER_MYSQL_CID" bash -c "cat \$(mysql -sse \"SELECT @@log_error;\")" > "$OUT_DIR/mysqld_error.log" 2>/dev/null
    FOUND_ERR_LOG="$OUT_DIR/mysqld_error.log"
fi
if [ -z "$FOUND_ERR_LOG" ]; then
    err "未找到MySQL错误日志，请手动指定日志路径"
    print_cmd "find / -name \"*.err\" 2>/dev/null"
else
    cp "$FOUND_ERR_LOG" "$OUT_DIR/mysqld_error.log"
    print_copy "MySQL错误日志完整路径" "$FOUND_ERR_LOG"
    echo "错误日志源文件: $FOUND_ERR_LOG" | tee -a "$OUT_DIR/report.log"
    # 提取致命错误、崩溃标记
    echo "--- 最近致命错误、崩溃警告 ---" >> "$OUT_DIR/report.log"
    grep -iE "error|fatal|assert|corrupt|tablespace|disk full|oom|signal 11|signal 9|permission denied" "$OUT_DIR/mysqld_error.log" | tail -30 | tee -a "$OUT_DIR/report.log"
fi
# 本体论分层分析注释（保留原有哲学逻辑）
# 1. Disk is full → 磁盘耗尽，优先清理binlog
# 2. InnoDB tablespace corrupted → 存储介质损坏，逐级innodb_force_recovery导出数据
# 3. OOM / out of memory → 进程被系统强制杀死，非程序主动退出
# 4. Too many open files → ulimit与my.cnf文件句柄配置不匹配
# 5. Signal 11(SIGSEGV) / Signal6(ABRT) → 内存越界/数据库内部断言崩溃
# ==========================================================================================================
# 步骤3：整体论关联——多维时间点交叉验证OOM/IO突变【单段可复制终端执行】
# ==========================================================================================================
section "3. 整体论交叉校验：OOM内存耗尽 / 磁盘IO硬件故障"
# 提取内核OOM、IO报错
DMESG_OOM=$(cat "$OUT_DIR/dmesg_oom_io_err.log" | grep -i "out of memory\|oom-killer" | tail -5)
DMESG_IO=$(cat "$OUT_DIR/dmesg_oom_io_err.log" | grep -i "i/o error\|read-only\|filesystem error\|blk dev" | tail -5)
# OOM内存耗尽链路校验
if [ -n "$DMESG_OOM" ]; then
    warn "【辩证矛盾点】检测到OOM Killer杀死进程，优先核对内存参数，而非单纯加内存"
    echo "$DMESG_OOM" | tee -a "$OUT_DIR/report.log"
    echo "--- 当前宿主机内存/swap快照 ---" >> "$OUT_DIR/report.log"
    free -h >> "$OUT_DIR/report.log"
    # 历史内存回溯（sysstat sar）
    if command -v sar &>/dev/null; then
        echo "--- 近1小时内存历史曲线 ---" >> "$OUT_DIR/report.log"
        sar -r -f /var/log/sa/sa$(date +%d) 2>/dev/null | grep -E "^[0-9]" | tail -12 >> "$OUT_DIR/report.log" || echo "当日sar数据不存在" >> "$OUT_DIR/report.log"
    else
        warn "未安装sysstat，无历史内存数据，手工命令：sar -r 1 10"
    fi
    print_cmd "grep innodb_buffer_pool_size $SELECTED_CNF"
fi
# IO/只读文件系统链路校验
if [ -n "$DMESG_IO" ]; then
    warn "【辩证矛盾点】磁盘IO异常/文件系统只读，会直接导致MySQL崩溃、写失败"
    echo "$DMESG_IO" | tee -a "$OUT_DIR/report.log"
    echo "--- 当前挂载列表，检查ro只读标记 ---" >> "$OUT_DIR/report.log"
    mount | grep -E "$DATA_DIR|/ " >> "$OUT_DIR/report.log"
    print_cmd "mount -o remount,rw $DATA_DIR挂载盘"
    print_cmd "fsck -n /dev/磁盘设备 # 停机后磁盘校验"
fi
# 整体论核心逻辑：对齐dmesg、错误日志、sar三者时间戳，识别连锁故障
# 示例因果链：内存耗尽OOM kill mysqld → 大量事务回滚 → binlog持续膨胀 → 磁盘100%占满
# 单一日志无法识别完整因果，必须多维度时间交叉比对
# ==========================================================================================================
# 步骤4：奥卡姆剃刀资源限制检查（磁盘、inode、文件句柄、binlog）【单段可复制终端执行】
# ==========================================================================================================
section "4. 资源硬限制排查：磁盘空间、inode、文件句柄、binlog堆积"
echo "--- 全局磁盘使用率（系统盘+MySQL数据盘+临时目录） ---" >> "$OUT_DIR/report.log"
df -h / "$DATA_DIR" /tmp 2>/dev/null | tee -a "$OUT_DIR/report.log"
echo "--- inode耗尽检查（小文件过多隐形崩溃） ---" >> "$OUT_DIR/report.log"
df -i / "$DATA_DIR" /tmp 2>/dev/null | tee -a "$OUT_DIR/report.log"
# binlog堆积检测（磁盘占满头号诱因）
if [ -d "$DATA_DIR" ]; then
    BINLOG_SIZE=$(du -sh "$DATA_DIR/mysql-bin."* 2>/dev/null | head -1)
    [ -n "$BINLOG_SIZE" ] && print_copy "Binlog总占用大小" "$BINLOG_SIZE" && echo "Binlog文件占用: $BINLOG_SIZE" | tee -a "$OUT_DIR/report.log"
fi
# 错误日志膨胀检查
if [ -f "$FOUND_ERR_LOG" ]; then
    ERR_LOG_SIZE=$(du -sh "$FOUND_ERR_LOG" 2>/dev/null | awk '{print $1}')
    print_copy "错误日志文件大小" "$ERR_LOG_SIZE"
    echo "错误日志大小: $ERR_LOG_SIZE (GB级日志需配置logrotate自动切割)" | tee -a "$OUT_DIR/report.log"
fi
# 文件句柄限制（too many open files崩溃根源）
echo "--- MySQL进程文件句柄硬限制 ---" >> "$OUT_DIR/report.log"
if [ -n "$MYSQL_PID" ]; then
    cat "/proc/$MYSQL_PID/limits" | grep "open files" | tee -a "$OUT_DIR/report.log"
    print_cmd "lsof -p $MYSQL_PID | wc -l # 统计当前打开文件总数"
else
    warn "无运行MySQL进程，无法读取进程级fd限制"
fi
echo "系统全局open file限制: $(ulimit -n)" >> "$OUT_DIR/report.log"
print_cmd "grep open_files_limit $SELECTED_CNF"
# 奥卡姆剃刀无为优化思路：
# 1. 配置binlog_expire_logs_seconds自动清理binlog，避免人工频繁清理
# 2. logrotate自动切割错误日志，防止日志无限膨胀占满磁盘
# 3. 统一调高系统与MySQL两层open_files_limit，消除句柄瓶颈
# ==========================================================================================================
# 步骤5：配置参数本体审查（内存、InnoDB核心参数校验）【单段可复制终端执行】
# ==========================================================================================================
section "5. MySQL核心配置文件审查，高危参数风险校验"
# 自动检索配置文件
for CNF in "${MYSQL_CNF_PATHS[@]}"; do
    if [ -f "$CNF" ]; then
        SELECTED_CNF="$CNF"
        cp "$CNF" "$OUT_DIR/my.cnf_backup"
        break
    fi
done
if [ -f "$SELECTED_CNF" ]; then
    print_copy "MySQL配置文件路径" "$SELECTED_CNF"
    echo "从 $SELECTED_CNF 提取风险核心参数:" | tee -a "$OUT_DIR/report.log"
    grep -E "innodb_buffer_pool_size|innodb_log_file_size|innodb_flush_log_at_trx_commit|max_connections|open_files_limit|binlog_expire|expire_logs_days|core-file" "$SELECTED_CNF" | tee -a "$OUT_DIR/report.log"
    # 本体论风险校验：buffer_pool超过物理内存70%极易触发OOM
    TOTAL_MEM=$(free -b | awk '/Mem:/ {print $2}')
    BUFF_POOL_RAW=$(grep innodb_buffer_pool_size "$SELECTED_CNF" | grep -oP '\d+' | head -1)
    if [ -n "$TOTAL_MEM" ] && [ -n "$BUFF_POOL_RAW" ]; then
        if [ "$BUFF_POOL_RAW" -gt $((TOTAL_MEM * 70 / 100)) ]; then
            err "高危风险：innodb_buffer_pool_size 超过物理内存70%，OOM崩溃概率极高，建议下调" | tee -a "$OUT_DIR/report.log"
        fi
    fi
else
    warn "未找到my.cnf配置文件，请手动查找配置"
    print_cmd "find /etc /usr/local -name my.cnf 2>/dev/null"
fi
# 补充SELinux权限校验（容器/宿主机通用权限崩溃根源）
section "5.1 SELinux权限审计（permission denied崩溃排查）"
ausearch -m avc -ts recent | grep -E mysql|mariadb > "$OUT_DIR/selinux_avc.log" 2>/dev/null
if [[ -s "$OUT_DIR/selinux_avc.log" ]]; then
    warn "检测到SELinux拦截MySQL文件读写，会导致启动失败/写入崩溃"
    cat "$OUT_DIR/selinux_avc.log"
    print_cmd "chcon -R -t mysqld_db_t $DATA_DIR"
    print_cmd "setsebool -P mysqld_disable_trans 1"
fi
# ==========================================================================================================
# 步骤6：辩证法恢复方案、无为长期预防、一键可复制修复命令【单段可复制终端执行】
# ==========================================================================================================
section "6. 辩证恢复决策清单 + 长期无为预防方案"
echo "===== 故障根因匹配恢复操作（全部命令预填充路径/PID，直接复制运行） =====" | tee -a "$OUT_DIR/report.log"
# 分场景自动匹配修复命令
if grep -qi "out of memory\|oom-killer" "$OUT_DIR/dmesg_oom_io_err.log"; then
    echo -e "\n【场景1：OOM内存耗尽被系统杀死】" | tee -a "$OUT_DIR/report.log"
    print_cmd "sed -i 's/innodb_buffer_pool_size=.*/innodb_buffer_pool_size=4G/' $SELECTED_CNF # 根据物理内存调整"
    print_cmd "echo 'vm.overcommit_memory=1' >> /etc/sysctl.conf && sysctl -p"
    print_cmd "systemctl restart mysqld"
    echo "短期缓解：降低buffer_pool；长期无为方案：cgroups限制MySQL内存、增加物理内存/swap" | tee -a "$OUT_DIR/report.log"
elif grep -qi "disk full" "$FOUND_ERR_LOG" 2>/dev/null; then
    echo -e "\n【场景2：磁盘空间耗尽】" | tee -a "$OUT_DIR/report.log"
    print_cmd "mysql -uroot -p -e \"PURGE BINARY LOGS BEFORE DATE(NOW() - INTERVAL 3 DAY);\""
    print_cmd "du -sh $DATA_DIR/* | sort -rh | head -10 # 定位大文件"
    echo "应急清理binlog；长期预防：配置binlog自动过期、磁盘使用率监控告警" | tee -a "$OUT_DIR/report.log"
elif grep -qi "corrupt\|tablespace\|assertion failed" "$OUT_DIR/mysqld_error.log"; then
    echo -e "\n【场景3：InnoDB表空间/数据文件损坏】" | tee -a "$OUT_DIR/report.log"
    print_cmd "sed -i 's/#innodb_force_recovery=.*/innodb_force_recovery=1/' $SELECTED_CNF"
    print_cmd "systemctl start mysqld && mysqldump -u root -p --all-databases > full_backup.sql"
    echo "辩证权衡：RPO优先→逐级恢复导出数据；RTO优先→直接使用备份重建实例" | tee -a "$OUT_DIR/report.log"
elif grep -qi "i/o error\|read-only file system" "$OUT_DIR/dmesg_oom_io_err.log"; then
    echo -e "\n【场景4：磁盘硬件IO故障/文件系统只读】" | tee -a "$OUT_DIR/report.log"
    print_cmd "mount -o remount,rw $DATA_DIR挂载分区"
    print_cmd "umount /data && fsck /dev/sdXX # 停机校验磁盘"
else
    echo -e "\n【无典型底层故障标记，执行全链路补充排查】" | tee -a "$OUT_DIR/report.log"
    print_cmd "ls -la $DATA_DIR # 核对目录属主mysql:mysql"
    print_cmd "ss -tlnp | grep 3306 # 检查端口占用"
    print_cmd "tail -n 100 $FOUND_ERR_LOG # 人工精读完整错误日志"
fi
# 整体论全链路扩展排查清单
echo -e "\n===== 全链路补充手工排查（上下游关联） =====" | tee -a "$OUT_DIR/report.log"
echo "1. 内核硬件层：dmesg -T 查看硬盘、内存硬件报错"
echo "2. 系统守护层：journalctl -u mysqld 完整启停日志"
echo "3. 存储层：检查RAID/云盘磁盘告警、磁盘读写延迟"
echo "4. 业务层：批量写入、大事务、无索引慢SQL触发资源暴涨"
echo "5. 集群层（主从）：SHOW SLAVE STATUS 复制线程崩溃、延迟堆积"
echo "6. 备份层：核对定时备份有效性，损坏场景用于数据恢复"
# 最终汇总输出
echo -e "\n${GREEN}==========================================${NC}"
info "MySQL系统化诊断全部采集完成"
print_copy "完整诊断存档目录" "$OUT_DIR"
print_copy "一键查看所有PID/路径清单命令" "cat $KEY_COPY_FILE"
warn "故障处理优先级建议：磁盘满 > OOM内存耗尽 > 数据文件损坏 > IO硬件故障 > 权限/配置错误"
echo -e "${GREEN}==========================================${NC}"

#!/bin/bash
# ============================================================================
# MySQL 故障系统化诊断脚本
# 用途：收集 MySQL 运行状态、配置、日志、系统资源等关键信息，用于故障排查
# 适用：Linux 系统，需有 MySQL 客户端访问权限（或本地 socket）
# 作者：DBA Team
# 版本：2026.07
# ============================================================================
# 设置输出目录
DIAG_DIR="/tmp/mysql_diag_$(date +%Y%m%d_%H%M%S)"
mkdir -p "$DIAG_DIR"
# 定义 MySQL 连接参数（建议使用 .my.cnf 或环境变量，避免密码明文）
# 若未设置，默认使用 unix socket 和当前系统用户
MYSQL_CMD="mysql -S /var/lib/mysql/mysql.sock -e"
# 若需远程连接，请修改为：
# MYSQL_CMD="mysql -h 127.0.0.1 -P 3306 -u diag_user -pdiag_pass -e"
# 定义日志文件
DIAG_LOG="$DIAG_DIR/diagnostic.log"
exec > >(tee -a "$DIAG_LOG") 2>&1
echo "============================================================"
echo "MySQL 诊断信息收集开始于: $(date)"
echo "诊断输出目录: $DIAG_DIR"
echo "============================================================"
# 1. 系统基本信息
echo ">>> 1. 系统基本信息"
uname -a
cat /etc/os-release 2>/dev/null || cat /etc/redhat-release 2>/dev/null
uptime
free -h
df -h
echo ""
# 2. MySQL 版本及运行状态
echo ">>> 2. MySQL 版本及运行状态"
$MYSQL_CMD "SELECT VERSION();"
$MYSQL_CMD "SHOW STATUS LIKE 'uptime';"
$MYSQL_CMD "SHOW VARIABLES LIKE 'port';"
$MYSQL_CMD "SHOW VARIABLES LIKE 'datadir';"
echo ""
# 3. MySQL 配置文件（过滤注释和空行）
echo ">>> 3. MySQL 配置文件内容"
if [ -f /etc/my.cnf ]; then
    grep -v '^#' /etc/my.cnf | grep -v '^$' > "$DIAG_DIR/my.cnf"
    cat "$DIAG_DIR/my.cnf"
elif [ -f /etc/mysql/my.cnf ]; then
    grep -v '^#' /etc/mysql/my.cnf | grep -v '^$' > "$DIAG_DIR/my.cnf"
    cat "$DIAG_DIR/my.cnf"
else
    echo "未找到 my.cnf"
fi
echo ""
# 4. 全局状态变量（关键指标）
echo ">>> 4. 关键状态变量"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Threads_connected';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Threads_running';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Max_used_connections';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Innodb_buffer_pool_reads';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Innodb_buffer_pool_read_requests';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Innodb_row_lock_waits';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Slow_queries';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Questions';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Com_select';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Com_insert';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Com_update';"
$MYSQL_CMD "SHOW GLOBAL STATUS LIKE 'Com_delete';"
echo ""
# 5. 当前进程列表（过滤用户和状态）
echo ">>> 5. 当前进程列表（前50个）"
$MYSQL_CMD "SHOW PROCESSLIST;" | head -n 50 > "$DIAG_DIR/processlist.txt"
cat "$DIAG_DIR/processlist.txt"
echo ""
# 6. InnoDB 引擎状态（关键信息）
echo ">>> 6. InnoDB 引擎状态（部分）"
$MYSQL_CMD "SHOW ENGINE INNODB STATUS\G" > "$DIAG_DIR/innodb_status.txt"
# 提取关键部分：锁等待、事务、I/O
grep -A 10 "LATEST DETECTED DEADLOCK" "$DIAG_DIR/innodb_status.txt"
grep -A 20 "TRANSACTIONS" "$DIAG_DIR/innodb_status.txt" | head -n 30
grep -A 10 "FILE I/O" "$DIAG_DIR/innodb_status.txt"
echo ""
# 7. 复制状态（如果启用）
echo ">>> 7. 复制状态"
$MYSQL_CMD "SHOW SLAVE STATUS\G" > "$DIAG_DIR/slave_status.txt"
if [ -s "$DIAG_DIR/slave_status.txt" ]; then
    grep -E "Slave_IO_Running|Slave_SQL_Running|Seconds_Behind_Master|Last_Err" "$DIAG_DIR/slave_status.txt"
else
    echo "未启用复制或无法获取复制状态"
fi
echo ""
# 8. 慢查询日志（最近50行）
echo ">>> 8. 慢查询日志（最近50行）"
SLOW_LOG=$(mysql -N -e "SHOW VARIABLES LIKE 'slow_query_log_file';" | awk '{print $2}')
if [ -f "$SLOW_LOG" ]; then
    tail -n 50 "$SLOW_LOG" > "$DIAG_DIR/slow_query.log"
    cat "$DIAG_DIR/slow_query.log"
else
    echo "未找到慢查询日志文件，请确认 slow_query_log 已开启"
fi
echo ""
# 9. 错误日志（最近100行）
echo ">>> 9. MySQL 错误日志（最近100行）"
ERROR_LOG=$(mysql -N -e "SHOW VARIABLES LIKE 'log_error';" | awk '{print $2}')
if [ -f "$ERROR_LOG" ]; then
    tail -n 100 "$ERROR_LOG" > "$DIAG_DIR/error.log"
    cat "$DIAG_DIR/error.log"
else
    echo "未找到错误日志文件"
fi
echo ""
# 10. 系统资源使用情况（MySQL 相关）
echo ">>> 10. 系统资源使用"
ps aux | grep -E 'mysqld|mysql' > "$DIAG_DIR/mysql_processes.txt"
cat "$DIAG_DIR/mysql_processes.txt"
echo ""
iostat -x 1 3 > "$DIAG_DIR/iostat.txt" 2>/dev/null
cat "$DIAG_DIR/iostat.txt"
echo ""
# 11. 磁盘 I/O 和文件系统信息
echo ">>> 11. 数据目录磁盘使用"
datadir=$(mysql -N -e "SHOW VARIABLES LIKE 'datadir';" | awk '{print $2}')
if [ -d "$datadir" ]; then
    du -sh "$datadir" 2>/dev/null
    ls -la "$datadir" | head -n 20
fi
echo ""
# 12. 网络连接状态（MySQL 端口）
echo ">>> 12. MySQL 端口连接数"
ss -tunap | grep -c 3306
ss -tunap | grep 3306 | head -n 20
echo ""
# 13. 最后，打包所有收集的文件
echo "============================================================"
echo "诊断信息收集完成，输出目录: $DIAG_DIR"
tar -czf "$DIAG_DIR.tar.gz" -C /tmp "$(basename "$DIAG_DIR")"
echo "打包文件: $DIAG_DIR.tar.gz"
echo "请将打包文件发送给 DBA 进行分析。"
echo "============================================================"
# 退出前提示
echo "注意：部分信息可能包含敏感数据，请妥善保管。"
exit 0

chmod +x collect_mysql_diag.sh

./collect_mysql_diag.sh