#!/usr/bin/env bash # ============================================================================ # build_devdb.sh — 构建本地开发库 eip4dev # 用途:从客户库 5435 按分层口径 dump/restore,完全脱离 5435 日常开发 # 使用:bash tools/local-devdb/build_devdb.sh [选项] # # 选项: # --src-host HOST 源库主机 (默认: 192.168.20.68) # --src-port PORT 源库端口 (默认: 5435) # --src-user USER 源库用户 (默认: ceshi) # --src-pass PASS 源库密码 (默认: ceshi) # --src-db DB 源库名 (默认: postgres) # --src-schema SCHEMA 源 schema (默认: eip3) # --dst-host HOST 目标主机 (默认: localhost) # --dst-port PORT 目标端口 (默认: 5433) # --dst-user USER 目标用户 (默认: zhengwr) # --dst-db DB 目标库名 (默认: eip4dev) # --threshold MB 全量/schema-only 阈值,MB (默认: 500) # --tables-config FILE 全量必须表配置文件 (默认: sql/data-tables.txt) # --work-dir DIR dump 工作目录 (默认: /tmp/eip4dev_build) # --skip-dump 跳过 dump,直接用 --work-dir 中已有文件 restore # --drop-target 重建前先 DROP 目标库(幂等重建) # -h, --help 显示帮助 # # 安全守门:严禁连接 5435 上的 ceshi 库(只读库副本,结构不完整) # ============================================================================ set -euo pipefail SCRIPT_DIR="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)" LOG_PREFIX="[eip4dev]" START_TS=$(date +%s) # ---- 默认参数 ---- SRC_HOST="192.168.20.68" SRC_PORT="5435" SRC_USER="ceshi" SRC_PASS="ceshi" SRC_DB="postgres" SRC_SCHEMA="eip3" DST_HOST="localhost" DST_PORT="5433" DST_USER="zhengwr" DST_DB="eip4dev" THRESHOLD_MB=500 TABLES_CONFIG="${SCRIPT_DIR}/sql/data-tables.txt" WORK_DIR="/tmp/eip4dev_build" SKIP_DUMP=0 DROP_TARGET=0 # ---- 参数解析 ---- while [[ $# -gt 0 ]]; do case "$1" in --src-host) SRC_HOST="$2"; shift 2 ;; --src-port) SRC_PORT="$2"; shift 2 ;; --src-user) SRC_USER="$2"; shift 2 ;; --src-pass) SRC_PASS="$2"; shift 2 ;; --src-db) SRC_DB="$2"; shift 2 ;; --src-schema) SRC_SCHEMA="$2"; shift 2 ;; --dst-host) DST_HOST="$2"; shift 2 ;; --dst-port) DST_PORT="$2"; shift 2 ;; --dst-user) DST_USER="$2"; shift 2 ;; --dst-db) DST_DB="$2"; shift 2 ;; --threshold) THRESHOLD_MB="$2"; shift 2 ;; --tables-config) TABLES_CONFIG="$2"; shift 2 ;; --work-dir) WORK_DIR="$2"; shift 2 ;; --skip-dump) SKIP_DUMP=1; shift ;; --drop-target) DROP_TARGET=1; shift ;; -h|--help) sed -n '2,40p' "$0" | grep '^#' | sed 's/^# \{0,1\}//' exit 0 ;; *) echo "未知参数: $1"; exit 1 ;; esac done # ---- 工具函数 ---- log() { echo "${LOG_PREFIX} $(date '+%H:%M:%S') $*"; } die() { echo "${LOG_PREFIX} ERROR: $*" >&2; exit 1; } elapsed() { echo $(( $(date +%s) - START_TS ))s; } # ---- 安全守门:严禁连接 5435 上的 ceshi 库 ---- if [[ "$SRC_HOST" == "192.168.20.68" && "$SRC_PORT" == "5435" && "$SRC_DB" == "ceshi" ]]; then die "安全守门触发:严禁连接 5435 上的 ceshi 库(数据不完整)!目标库应为 postgres(currentSchema=eip3)" fi export PGPASSWORD="$SRC_PASS" SRC_CONN="-h $SRC_HOST -p $SRC_PORT -U $SRC_USER -d $SRC_DB" DST_CONN="-h $DST_HOST -p $DST_PORT -U $DST_USER" # ---- 检查依赖工具 ---- for tool in pg_dump pg_restore psql; do command -v "$tool" > /dev/null 2>&1 || die "缺少工具: ${tool} (需要 PG15 客户端)" done PG_VER=$(pg_dump --version | awk '{print $3}' | cut -d. -f1) log "pg_dump version: $PG_VER (need >= 15)" [[ "$PG_VER" -ge 15 ]] || die "pg_dump version too old, need PG15 client" # ---- 验证源库连接 ---- log "验证源库连接 ${SRC_HOST}:${SRC_PORT}/${SRC_DB}..." SRC_PG_VER=$(PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc "SELECT version();" 2>&1 | head -1) || die "源库连接失败" log "源库: $SRC_PG_VER" # ---- 统计源库分层 ---- log "统计源库表规模分层..." TOTAL_TABLES=$(PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc \ "SELECT COUNT(*) FROM pg_stat_user_tables WHERE schemaname='${SRC_SCHEMA}';") ACT_TABLES=$(PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc \ "SELECT COUNT(*) FROM pg_stat_user_tables WHERE schemaname='${SRC_SCHEMA}' AND relname LIKE 'act_%';") GIANT_TABLES=$(PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc \ "SELECT COUNT(*) FROM pg_stat_user_tables WHERE schemaname='${SRC_SCHEMA}' AND relname NOT LIKE 'act_%' AND pg_total_relation_size(quote_ident(schemaname)||'.'||quote_ident(relname)) > ${THRESHOLD_MB}*1024*1024;") FULLDATA_TABLES=$(( TOTAL_TABLES - ACT_TABLES - GIANT_TABLES )) log "表统计:总计 ${TOTAL_TABLES} / ACT_*(schema-only) ${ACT_TABLES} / 超${THRESHOLD_MB}MB巨表(schema-only) ${GIANT_TABLES} / 全量数据表 ≈${FULLDATA_TABLES}" # ---- 生成巨表名单(非ACT_,超阈值)---- GIANT_LIST_FILE="${WORK_DIR}/giant_tables.txt" mkdir -p "$WORK_DIR" PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc \ "SELECT relname FROM pg_stat_user_tables WHERE schemaname='${SRC_SCHEMA}' AND relname NOT LIKE 'act_%' AND pg_total_relation_size(quote_ident(schemaname)||'.'||quote_ident(relname)) > ${THRESHOLD_MB}*1024*1024 ORDER BY pg_total_relation_size(quote_ident(schemaname)||'.'||quote_ident(relname)) DESC;" > "$GIANT_LIST_FILE" log "超阈值巨表清单 (${GIANT_LIST_FILE}):" while IFS= read -r t; do SZ=$(PGPASSWORD="$SRC_PASS" psql $SRC_CONN -tAc \ "SELECT pg_size_pretty(pg_total_relation_size('${SRC_SCHEMA}.${t}'));") log " ${t}: ${SZ}" done < "$GIANT_LIST_FILE" # ---- 准备工作目录 ---- SCHEMA_DUMP="${WORK_DIR}/schema.dump" DATA_DUMP="${WORK_DIR}/data.dump" log "工作目录: ${WORK_DIR}" if [[ $SKIP_DUMP -eq 0 ]]; then # ---- Step 1: Schema-only dump(全 schema,所有表 DDL)---- log "=== Step 1/4: Schema-only dump 开始 ===" T1=$(date +%s) PGPASSWORD="$SRC_PASS" pg_dump \ -h "$SRC_HOST" -p "$SRC_PORT" -U "$SRC_USER" -d "$SRC_DB" \ -n "$SRC_SCHEMA" \ --schema-only \ -Fc \ -f "$SCHEMA_DUMP" T2=$(date +%s) SCHEMA_SZ=$(du -sh "$SCHEMA_DUMP" | cut -f1) log "Schema dump 完成: ${SCHEMA_SZ},耗时 $((T2-T1))s" # ---- Step 2: 全量数据 dump(排除 ACT_* 和超阈值巨表)---- log "=== Step 2/4: 数据 dump 开始(排除 ACT_* 和超阈值巨表)===" T1=$(date +%s) # 构建 --exclude-table-data 参数:ACT_* 全排 EXCLUDE_ARGS=() EXCLUDE_ARGS+=(--exclude-table-data="${SRC_SCHEMA}.act_*") # 排除超阈值巨表数据 while IFS= read -r tbl; do [[ -n "$tbl" ]] && EXCLUDE_ARGS+=(--exclude-table-data="${SRC_SCHEMA}.${tbl}") done < "$GIANT_LIST_FILE" PGPASSWORD="$SRC_PASS" pg_dump \ -h "$SRC_HOST" -p "$SRC_PORT" -U "$SRC_USER" -d "$SRC_DB" \ -n "$SRC_SCHEMA" \ --data-only \ "${EXCLUDE_ARGS[@]}" \ -Fc \ -f "$DATA_DUMP" T2=$(date +%s) DATA_SZ=$(du -sh "$DATA_DUMP" | cut -f1) log "数据 dump 完成: ${DATA_SZ},耗时 $((T2-T1))s" else log "跳过 dump(--skip-dump 模式),使用已有文件: schema=${SCHEMA_DUMP} data=${DATA_DUMP}" [[ -f "$SCHEMA_DUMP" ]] || die "schema dump 文件不存在: $SCHEMA_DUMP" [[ -f "$DATA_DUMP" ]] || die "data dump 文件不存在: $DATA_DUMP" fi # ---- Step 3: 准备目标库 ---- log "=== Step 3/4: 准备目标库 ${DST_DB} ===" unset PGPASSWORD if [[ $DROP_TARGET -eq 1 ]]; then log "DROP 并重建目标库 ${DST_DB}..." psql $DST_CONN -d postgres -c "DROP DATABASE IF EXISTS ${DST_DB};" 2>&1 | grep -v "^$" || true fi # 创建库(已存在则跳过) psql $DST_CONN -d postgres -c "SELECT 1 FROM pg_database WHERE datname='${DST_DB}';" 2>&1 | grep -q "1 row" \ || psql $DST_CONN -d postgres -c "CREATE DATABASE ${DST_DB} ENCODING 'UTF8';" log "目标库 ${DST_DB} 就绪" # 建角色(P0' 口径:restore 需要 eip3/ceshi 角色存在) psql $DST_CONN -d "$DST_DB" -c " DO \$\$ BEGIN IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname='eip3') THEN CREATE ROLE eip3; RAISE NOTICE 'role eip3 created'; ELSE RAISE NOTICE 'role eip3 already exists'; END IF; END\$\$;" 2>&1 | grep -v "^$" psql $DST_CONN -d "$DST_DB" -c " DO \$\$ BEGIN IF NOT EXISTS (SELECT 1 FROM pg_roles WHERE rolname='ceshi') THEN CREATE ROLE ceshi LOGIN; RAISE NOTICE 'role ceshi created'; ELSE RAISE NOTICE 'role ceshi already exists'; END IF; END\$\$;" 2>&1 | grep -v "^$" # 安装 uuid-ossp 扩展(11 个函数默认值依赖) psql $DST_CONN -d "$DST_DB" -c "CREATE EXTENSION IF NOT EXISTS \"uuid-ossp\";" 2>&1 | grep -v "^$" log "前置角色和扩展就绪" # ---- Step 4: Restore ---- log "=== Step 4/4: Restore 到 ${DST_DB} ===" # 4a: Schema restore T1=$(date +%s) log "4a: restore schema..." pg_restore \ $DST_CONN \ -d "$DST_DB" \ --no-owner \ --no-privileges \ --disable-triggers \ -e \ "$SCHEMA_DUMP" 2>&1 | grep -v "^$" | grep -v "already exists" | head -50 || true T2=$(date +%s) log "Schema restore 完成,耗时 $((T2-T1))s" # 4b: Data restore(禁用触发器,避免 FK 顺序问题) T1=$(date +%s) log "4b: restore data(--disable-triggers)..." pg_restore \ $DST_CONN \ -d "$DST_DB" \ --no-owner \ --no-privileges \ --disable-triggers \ --single-transaction \ -e \ "$DATA_DUMP" 2>&1 | grep -v "^$" | head -100 || true T2=$(date +%s) log "数据 restore 完成,耗时 $((T2-T1))s" # ---- 验证与统计 ---- log "=== 验证与统计 ===" DST_TABLE_COUNT=$(psql $DST_CONN -d "$DST_DB" -tAc \ "SELECT COUNT(*) FROM pg_stat_user_tables WHERE schemaname='${SRC_SCHEMA}';") log "目标库表数: ${DST_TABLE_COUNT}(源库: ${TOTAL_TABLES})" # 关键引擎元数据行数验证 log "关键表行数验证:" for tbl in sys_entity sys_entityattribute sys_reference sys_referencevalue sys_user view_template view_page sys_serviceflow; do CNT=$(psql $DST_CONN -d "$DST_DB" -tAc "SELECT COUNT(*) FROM ${SRC_SCHEMA}.${tbl};" 2>/dev/null || echo "N/A") log " ${tbl}: ${CNT} 行" done # 目标库体量 DB_SIZE=$(psql $DST_CONN -d "$DST_DB" -tAc \ "SELECT pg_size_pretty(pg_database_size('${DST_DB}'));") log "目标库 ${DST_DB} 总体量: ${DB_SIZE}" TOTAL_ELAPSED=$(elapsed) log "=== 构建完成!总耗时: ${TOTAL_ELAPSED} ===" log "连接串: jdbc:postgresql://${DST_HOST}:${DST_PORT}/${DST_DB}?currentSchema=${SRC_SCHEMA}" log "用户: ${DST_USER}(无密码)" # ---- 分层汇总 ---- echo "" echo "┌─────────────────────────────────────────────────────┐" echo "│ eip4dev 构建完成 - 分层汇总 │" echo "├──────────────────────────┬──────────────────────────┤" printf "│ %-26s│ %-26s│\n" "总表数(DDL)" "${TOTAL_TABLES}" printf "│ %-26s│ %-26s│\n" "ACT_*(schema-only)" "${ACT_TABLES}" printf "│ %-26s│ %-26s│\n" "非ACT_巨表(schema-only)" "${GIANT_TABLES}" printf "│ %-26s│ %-26s│\n" "全量数据表 ≈" "${FULLDATA_TABLES}" printf "│ %-26s│ %-26s│\n" "目标库体量" "${DB_SIZE}" printf "│ %-26s│ %-26s│\n" "总耗时" "${TOTAL_ELAPSED}" echo "└──────────────────────────┴──────────────────────────┘"