Guard server provision nginx install

2026-05-02 20:25:01 +08:00
parent 64fda2a677
commit 7160e90909
2 changed files with 106 additions and 12 deletions
--- a/docs/technical/PRODUCTION_DEPLOYMENT_PLAN_2026-05-02.md
+++ b/docs/technical/PRODUCTION_DEPLOYMENT_PLAN_2026-05-02.md
@@ -350,12 +350,16 @@ WASM_SOURCE="${CARGO_TARGET_DIR}/wasm32-unknown-unknown/release/spacetime_module
 - 创建 `/stdb`、`/opt/genarrative`、`/srv/genarrative`、`/etc/genarrative`、`/var/lib/genarrative/maintenance`。
 - 安装或更新 SpacetimeDB。
 - 安装 systemd unit。
- 安装 Nginx 配置和维护模式 snippet。
- 执行 `nginx -t`。
+- 可选安装 Nginx 配置和维护模式 snippet。
+- 安装 Nginx 配置时执行 `nginx -t`。
 - 启用并启动 `spacetimedb.service` 与 `genarrative-api.service`。

 该流水线属于高风险操作，默认要求人工确认后执行。
-已落地的 Jenkinsfile 为 `jenkins/Jenkinsfile.production-server-provision`。该流水线默认 `DRY_RUN=true`，只打印将执行的初始化动作；真正写入系统用户、目录、systemd、Nginx 配置并启动服务时，必须设置 `DRY_RUN=false` 且勾选 `CONFIRM_PROVISION`。当 `DEPLOY_TARGET=release` 时，还必须勾选 `CONFIRM_RELEASE_DEPLOY_AGENT`，并通过 `linux && genarrative-release-deploy` 调度到独立 release 部署 agent。
+已落地的 Jenkinsfile 为 `jenkins/Jenkinsfile.production-server-provision`。该流水线默认 `DRY_RUN=true`，只打印将执行的初始化动作；真正写入系统用户、目录、systemd、环境文件并启动服务时，必须设置 `DRY_RUN=false` 且勾选 `CONFIRM_PROVISION`。当 `DEPLOY_TARGET=release` 时，还必须勾选 `CONFIRM_RELEASE_DEPLOY_AGENT`，并通过 `linux && genarrative-release-deploy` 调度到独立 release 部署 agent。
+
+首次真实初始化默认保持 `INSTALL_NGINX_CONFIG=false`，先完成系统用户、目录、SpacetimeDB、systemd unit 与 `/etc/genarrative/api-server.env` 落盘。等真实域名确定，并且目标机已经存在 `/etc/letsencrypt/live/<SERVER_NAME>/fullchain.pem` 与 `/etc/letsencrypt/live/<SERVER_NAME>/privkey.pem` 后，再把 `SERVER_NAME` 从 `genarrative.example.com` 改成真实域名，并设置 `INSTALL_NGINX_CONFIG=true` 安装 Nginx HTTPS 配置。流水线会拒绝在真实初始化中用占位域名写入 Nginx 配置，也会在证书缺失时提前失败。
+
+若误用占位域名执行过真实初始化，失败通常发生在 `nginx -t`，错误表现为找不到 `/etc/letsencrypt/live/genarrative.example.com/fullchain.pem` 或 `privkey.pem`。新版初始化在 `INSTALL_NGINX_CONFIG=false` 时会检测并禁用上一轮留下的占位域名 Nginx 配置，避免它继续影响后续 `nginx -t`。

 ### Web Build / Deploy

--- a/jenkins/Jenkinsfile.production-server-provision
+++ b/jenkins/Jenkinsfile.production-server-provision
@@ -26,7 +26,7 @@ pipeline {
    string(name: 'WEB_LINK', defaultValue: '/srv/genarrative/web', description: 'Nginx 静态站点目录或软链接')
    string(name: 'API_ENV_FILE', defaultValue: '/etc/genarrative/api-server.env', description: 'api-server 环境文件')
    string(name: 'API_PORT', defaultValue: '8082', description: 'api-server 本机监听端口')
-    booleanParam(name: 'INSTALL_NGINX_CONFIG', defaultValue: true, description: '安装 Nginx 配置并执行 nginx -t')
+    booleanParam(name: 'INSTALL_NGINX_CONFIG', defaultValue: false, description: '安装 Nginx 配置并执行 nginx -t；首次初始化且证书未准备好时保持关闭')
    booleanParam(name: 'ENABLE_SERVICES', defaultValue: true, description: '启用并启动 spacetimedb 与 api-server systemd 服务')
  }

@@ -49,6 +49,9 @@ pipeline {
          if (!params.SPACETIME_BIN_SOURCE?.trim()) {
            error('SPACETIME_BIN_SOURCE 不能为空。')
          }
+          if (!params.DRY_RUN && params.INSTALL_NGINX_CONFIG && params.SERVER_NAME?.trim() == 'genarrative.example.com') {
+            error('真实初始化安装 Nginx 配置时必须把 SERVER_NAME 改成真实域名，不能使用 genarrative.example.com 占位值。证书未准备好时请先保持 INSTALL_NGINX_CONFIG=false。')
+          }
        }
      }
    }
@@ -124,6 +127,98 @@ pipeline {
                deploy/env/api-server.env.example
            }

+            validate_nginx_tls() {
+              local cert_dir="/etc/letsencrypt/live/${SERVER_NAME}"
+              if [[ "${SERVER_NAME}" == "genarrative.example.com" ]]; then
+                echo "[server-provision] SERVER_NAME 仍是占位域名，拒绝写入 Nginx HTTPS 配置。请填写真实域名，或先设置 INSTALL_NGINX_CONFIG=false。" >&2
+                exit 1
+              fi
+              if [[ ! -f "${cert_dir}/fullchain.pem" || ! -f "${cert_dir}/privkey.pem" ]]; then
+                echo "[server-provision] 未找到 Nginx HTTPS 证书: ${cert_dir}/fullchain.pem 或 ${cert_dir}/privkey.pem" >&2
+                echo "[server-provision] 请先完成证书申请，或首次初始化时设置 INSTALL_NGINX_CONFIG=false，避免写入无法通过 nginx -t 的配置。" >&2
+                exit 1
+              fi
+            }
+
+            install_nginx_config_with_rollback() {
+              local config_target="/etc/nginx/conf.d/genarrative.conf"
+              local snippet_target="/etc/nginx/snippets/genarrative-maintenance.conf"
+              local rendered_config rendered_snippet config_backup snippet_backup
+              local had_config="false"
+              local had_snippet="false"
+
+              run_cmd mkdir -p /etc/nginx/snippets /etc/nginx/conf.d
+              echo "+ render deploy/nginx/genarrative.conf -> ${config_target}"
+              echo "+ install -m 0644 deploy/nginx/snippets/genarrative-maintenance.conf ${snippet_target}"
+
+              if [[ "${DRY_RUN}" == "true" ]]; then
+                echo "+ nginx -t"
+                return
+              fi
+
+              validate_nginx_tls
+              rendered_config="$(mktemp)"
+              rendered_snippet="$(mktemp)"
+              config_backup="$(mktemp)"
+              snippet_backup="$(mktemp)"
+              render_nginx_config >"${rendered_config}"
+              cp deploy/nginx/snippets/genarrative-maintenance.conf "${rendered_snippet}"
+
+              if [[ -f "${config_target}" ]]; then
+                cp -p "${config_target}" "${config_backup}"
+                had_config="true"
+              fi
+              if [[ -f "${snippet_target}" ]]; then
+                cp -p "${snippet_target}" "${snippet_backup}"
+                had_snippet="true"
+              fi
+
+              install -m 0644 "${rendered_config}" "${config_target}"
+              install -m 0644 "${rendered_snippet}" "${snippet_target}"
+
+              if ! nginx -t; then
+                echo "[server-provision] nginx -t 失败，恢复写入前的 Nginx 配置。" >&2
+                if [[ "${had_config}" == "true" ]]; then
+                  cp -p "${config_backup}" "${config_target}"
+                else
+                  rm -f "${config_target}"
+                fi
+                if [[ "${had_snippet}" == "true" ]]; then
+                  cp -p "${snippet_backup}" "${snippet_target}"
+                else
+                  rm -f "${snippet_target}"
+                fi
+                rm -f "${rendered_config}" "${rendered_snippet}" "${config_backup}" "${snippet_backup}"
+                exit 1
+              fi
+
+              rm -f "${rendered_config}" "${rendered_snippet}" "${config_backup}" "${snippet_backup}"
+            }
+
+            cleanup_placeholder_nginx_config() {
+              local config_target="/etc/nginx/conf.d/genarrative.conf"
+              local disabled_target
+
+              if [[ ! -f "${config_target}" ]]; then
+                return
+              fi
+
+              if ! grep -q "/etc/letsencrypt/live/genarrative.example.com/" "${config_target}"; then
+                return
+              fi
+
+              disabled_target="${config_target}.disabled-placeholder-$(date +%Y%m%d%H%M%S)"
+              echo "[server-provision] 发现上一轮初始化留下的占位域名 Nginx 配置，禁用: ${config_target} -> ${disabled_target}"
+              if [[ "${DRY_RUN}" != "true" ]]; then
+                mv "${config_target}" "${disabled_target}"
+                if command -v nginx >/dev/null 2>&1; then
+                  if ! nginx -t; then
+                    echo "[server-provision] 占位配置已禁用，但 nginx -t 仍失败；请检查其他 Nginx 配置。" >&2
+                  fi
+                fi
+              fi
+            }
+
            escape_sed_replacement() {
              printf "%s" "$1" | sed "s/[&|]/\\\\&/g"
            }
@@ -205,14 +300,9 @@ pipeline {
            fi

            if [[ "${INSTALL_NGINX_CONFIG}" == "true" ]]; then
-              run_cmd mkdir -p /etc/nginx/snippets /etc/nginx/conf.d
-              echo "+ render deploy/nginx/genarrative.conf -> /etc/nginx/conf.d/genarrative.conf"
-              if [[ "${DRY_RUN}" != "true" ]]; then
-                render_nginx_config >/etc/nginx/conf.d/genarrative.conf
-                chmod 0644 /etc/nginx/conf.d/genarrative.conf
-              fi
-              install_file deploy/nginx/snippets/genarrative-maintenance.conf /etc/nginx/snippets/genarrative-maintenance.conf 0644
-              run_cmd nginx -t
+              install_nginx_config_with_rollback
+            else
+              cleanup_placeholder_nginx_config
            fi

            run_cmd systemctl daemon-reload