lzh
8c5c5ef44a
- 新增 Pre-deploy Check：SSH 到 Prod/Registry 读根分区空闲，<5% 直接 fail（规避磁盘满时 sshd 连带崩溃导致的 scp 失败），5~10% 仅告警 - 新增 Cleanup Old Images stage：部署成功后每服务保留最近 3 个镜像 * Prod 侧调用 scripts/cleanup.sh * Registry 侧调用 scripts/registry-cleanup.py + 触发容器内 garbage-collect - scripts/cleanup.sh：去掉 volume prune 的交互 read（CI 下会卡住），支持 --keep/--prune-volumes/--registry 参数 - scripts/registry-cleanup.py：按 tag 内数字降序保留最新 N 个；覆盖 Docker v2/OCI 多种 manifest Accept；多 tag 指向同一 digest 去重；失败不影响发布 Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
124 lines
3.3 KiB
Bash
124 lines
3.3 KiB
Bash
#!/bin/bash
|
||
|
||
# ============================================
|
||
# AIOT Platform - 清理脚本
|
||
# 清理部署主机上的旧镜像 / 停止容器 / 构建缓存，释放存储空间
|
||
# ============================================
|
||
|
||
set -e
|
||
|
||
# ---- 默认参数 ----
|
||
KEEP=3
|
||
PRUNE_VOLUMES=false
|
||
REGISTRY_HOST="localhost:5000"
|
||
|
||
usage() {
|
||
cat <<EOF
|
||
用法: $0 [options]
|
||
|
||
Options:
|
||
--keep=N 每个服务保留最近 N 个本地镜像（默认 3）
|
||
--prune-volumes 额外清理未使用的 Docker volume（默认不清，避免误删数据）
|
||
--registry=HOST 本地 docker images 仓库前缀（默认 localhost:5000）
|
||
-h, --help 帮助
|
||
|
||
示例：
|
||
$0 --keep=3
|
||
$0 --keep=2 --prune-volumes
|
||
EOF
|
||
}
|
||
|
||
for arg in "$@"; do
|
||
case "$arg" in
|
||
--keep=*) KEEP="${arg#*=}" ;;
|
||
--prune-volumes) PRUNE_VOLUMES=true ;;
|
||
--registry=*) REGISTRY_HOST="${arg#*=}" ;;
|
||
-h|--help) usage; exit 0 ;;
|
||
*) echo "未知参数: $arg"; usage; exit 1 ;;
|
||
esac
|
||
done
|
||
|
||
if ! [[ "$KEEP" =~ ^[0-9]+$ ]] || [ "$KEEP" -lt 1 ]; then
|
||
echo "❌ --keep 必须为正整数"
|
||
exit 1
|
||
fi
|
||
|
||
# 颜色输出
|
||
GREEN='\033[0;32m'
|
||
YELLOW='\033[1;33m'
|
||
NC='\033[0m'
|
||
|
||
log_info() { echo -e "${GREEN}[INFO]${NC} $1"; }
|
||
log_warn() { echo -e "${YELLOW}[WARN]${NC} $1"; }
|
||
|
||
log_info "========================================="
|
||
log_info "AIOT Platform 清理开始 (keep=${KEEP})"
|
||
log_info "========================================="
|
||
|
||
log_info "当前磁盘使用情况:"
|
||
df -h | grep -E "Filesystem|/$" || true
|
||
echo ""
|
||
log_info "当前 Docker 磁盘使用:"
|
||
docker system df
|
||
echo ""
|
||
|
||
log_info "清理停止的容器..."
|
||
docker container prune -f
|
||
|
||
log_info "清理悬空镜像..."
|
||
docker image prune -f
|
||
|
||
log_info "清理旧版本镜像（每个服务保留最近 ${KEEP} 个）..."
|
||
|
||
SERVICES=(
|
||
"viewsh-gateway"
|
||
"viewsh-module-system-server"
|
||
"viewsh-module-infra-server"
|
||
"viewsh-module-iot-server"
|
||
"viewsh-module-iot-gateway"
|
||
"viewsh-module-ops-server"
|
||
)
|
||
|
||
for service in "${SERVICES[@]}"; do
|
||
log_info "处理服务: ${service}"
|
||
|
||
# 按创建时间降序取 ID 列表，跳过 latest tag，保留前 KEEP 个
|
||
mapfile -t ids_to_delete < <(
|
||
docker images "${REGISTRY_HOST}/${service}" \
|
||
--format '{{.CreatedAt}}|{{.ID}}|{{.Tag}}' \
|
||
| grep -v '|latest$' \
|
||
| sort -r \
|
||
| awk -F'|' -v k="$KEEP" 'NR > k {print $2}'
|
||
)
|
||
|
||
if [ "${#ids_to_delete[@]}" -eq 0 ]; then
|
||
log_info " └─ 无可清理镜像"
|
||
continue
|
||
fi
|
||
|
||
log_info " └─ 删除 ${#ids_to_delete[@]} 个旧镜像"
|
||
# 去重后批量删
|
||
printf '%s\n' "${ids_to_delete[@]}" | sort -u | xargs -r docker rmi -f 2>/dev/null || true
|
||
done
|
||
|
||
if [ "$PRUNE_VOLUMES" = true ]; then
|
||
log_warn "清理未使用的 volume（--prune-volumes 已启用）"
|
||
docker volume prune -f
|
||
else
|
||
log_info "跳过 volume 清理（如需清理请加 --prune-volumes）"
|
||
fi
|
||
|
||
log_info "清理 Docker 构建缓存（24h 前）..."
|
||
docker builder prune -f --filter "until=24h" || true
|
||
|
||
echo ""
|
||
log_info "========================================="
|
||
log_info "清理完成"
|
||
log_info "========================================="
|
||
echo ""
|
||
log_info "清理后磁盘使用情况:"
|
||
df -h | grep -E "Filesystem|/$" || true
|
||
echo ""
|
||
log_info "清理后 Docker 磁盘使用:"
|
||
docker system df
|