diff --git a/agent/README.md b/agent/README.md index 796e3d6..252b2f4 100644 --- a/agent/README.md +++ b/agent/README.md @@ -164,4 +164,4 @@ cd agent make # CGO_ENABLED=0 静态编译当前平台 ``` -顶层 `./run.sh build agent` 会交叉编译 `linux-amd64`、`linux-arm64`、`darwin-arm64`、`windows-amd64.exe`。Docker 镜像会携带当前目标架构 Linux Agent 和 Windows AMD64 Agent。// Logan.Li +顶层 `./run.sh build agent` 会交叉编译 `linux-amd64`、`linux-arm64`、`windows-amd64.exe`。Docker 镜像内置同样的三个平台 Agent 二进制，覆盖主流服务器架构。// Logan.Li diff --git a/agent/deploy/install.sh b/agent/deploy/install.sh index 6451c58..2500739 100644 --- a/agent/deploy/install.sh +++ b/agent/deploy/install.sh @@ -106,7 +106,7 @@ curl -fSL -o /tmp/aiops-agent "${BINARY_URL}" || { echo "❌ 下载失败: ${BINARY_URL}" echo " 请检查: 1) 后端地址是否正确 2) 该平台是否有对应二进制" if [ "$OS_LOWER" = "darwin" ]; then - echo " 提示: macOS 二进制需要后端编译 darwin 版本 (make build-darwin-arm64)" + echo " 提示: 平台不再提供 macOS Agent 二进制，服务器采集请使用 Linux/Windows 主机" fi exit 1 } diff --git a/agent/main.go b/agent/main.go index 9e569cf..55b3c54 100644 --- a/agent/main.go +++ b/agent/main.go @@ -35,7 +35,7 @@ import ( ) var ( - version = "v2.1.0" + version = "v2.2.0" buildTime = "unknown" ) @@ -292,33 +292,92 @@ func main() { } } - const pollInterval = 5 * time.Second + // ── 配置轮询 + 上报循环（v2.2）──────────────────────────────── + // + // 旧版每 5 秒分别拉取 tasks/poll、health-probes/config、port-policy 三个接口， + // 配置类请求量是上报本身的 6 倍；现改为单次 GET /agent/config， + // 并由服务端下发 report_interval，Agent 热更新 ticker、无需重装。 // Logan.Li + const minInterval = 5 * time.Second + const fallbackConfigPollInterval = 30 * time.Second + intervalUpdates := make(chan time.Duration, 1) + + configTicker := time.NewTicker(fallbackConfigPollInterval) go func() { - pollAndExecute() - pullProbes() - pullPortPolicy() - ticker := time.NewTicker(pollInterval) - defer ticker.Stop() - for { - select { - case <-ticker.C: + defer configTicker.Stop() + runConfig := func() { + merged, err := taskClient.PollConfig(ctx) + if err != nil { + if ctx.Err() != nil { + return + } + // 服务端尚未升级（或暂时失败）时回退旧的分接口模式，保证兼容。 // Logan.Li pollAndExecute() pullProbes() pullPortPolicy() + return + } + + // 执行服务端下发的任务。 // Logan.Li + for _, t := range merged.Tasks { + if ctx.Err() != nil { + return + } + result := taskExecutor.ExecuteTask(ctx, t) + if err := taskClient.Submit(ctx, result); err != nil { + logger.Printf("任务 %s 结果回传失败: %v", t.TaskID, err) + } + } + if len(merged.HealthProbes) > 0 { + probeConfigs = merged.HealthProbes + } + if merged.PortPolicy != nil { + portCollector.SetPolicy(merged.PortPolicy) + } + + // 服务端下发的上报间隔 → 通知主循环热更新 ticker。 // Logan.Li + if merged.ReportInterval > 0 { + next := time.Duration(merged.ReportInterval) * time.Second + if next < minInterval { + next = minInterval + } + select { + case intervalUpdates <- next: + default: + } + } + // 服务端下发的配置轮询间隔 → 调整本循环节奏。 // Logan.Li + if merged.ConfigPollInterval > 0 { + next := time.Duration(merged.ConfigPollInterval) * time.Second + if next >= minInterval && next != fallbackConfigPollInterval { + configTicker.Reset(next) + } + } + } + + runConfig() + for { + select { + case <-configTicker.C: + runConfig() case <-ctx.Done(): return } } }() + // 上报循环：间隔可被服务端热更新。 // Logan.Li time.Sleep(startupJitter) collectAndReport() - ticker := time.NewTicker(time.Duration(cfg.Interval) * time.Second) - defer ticker.Stop() + reportTicker := time.NewTicker(time.Duration(cfg.Interval) * time.Second) + defer reportTicker.Stop() for { select { - case <-ticker.C: + case <-reportTicker.C: collectAndReport() + case next := <-intervalUpdates: + // 重新抖动，避免多台主机在服务端下发同值后被对齐成同一时刻上报。 // Logan.Li + reportTicker.Reset(next) + logger.Printf("服务端下发新的上报间隔: %s", next) case <-ctx.Done(): logger.Println("收到退出信号，Agent 正常关闭") return diff --git a/agent/task/client.go b/agent/task/client.go index 20d7753..8593df6 100644 --- a/agent/task/client.go +++ b/agent/task/client.go @@ -20,6 +20,8 @@ const ( resultPathFmt = "/api/v1/agent/tasks/%s/result" probePath = "/api/v1/agent/health-probes/config" portPolicyPath = "/api/v1/agent/port-policy" + // configPath 合并配置接口（v2.2）：一次拉取任务+探针+端口策略+采集频率。 // Logan.Li + configPath = "/api/v1/agent/config" ) // Client 负责与后端任务/探针接口通信。 @@ -90,3 +92,32 @@ func (c *Client) PollPortPolicy(ctx context.Context) (*model.PortPolicy, error) } return &wrapper.Data, nil } + +// AgentConfig 合并配置响应（v2.2）。 +// +// 旧版 Agent 每 5 秒要分别请求 tasks/poll、health-probes/config、port-policy +// 三个接口，配置类请求量是上报本身的 6 倍；合并后单次请求即可拿到全部配置。 // Logan.Li +type AgentConfig struct { + ConfigVersion string `json:"config_version"` + ReportInterval int `json:"report_interval"` + ConfigPollInterval int `json:"config_poll_interval"` + Tasks []Task `json:"tasks"` + HealthProbes []model.ProbeConfig `json:"health_probes"` + PortPolicy *model.PortPolicy `json:"port_policy,omitempty"` +} + +// PollConfig 一次性拉取任务 + 探针 + 端口策略 + 采集频率（v2.2 合并接口）。 +// 若服务端尚未升级（404），调用方应回退到分接口拉取。 // Logan.Li +func (c *Client) PollConfig(ctx context.Context) (*AgentConfig, error) { + path := configPath + "?hostname=" + url.QueryEscape(c.hostname) + + var wrapper struct { + Code int `json:"code"` + Message string `json:"message"` + Data AgentConfig `json:"data"` + } + if err := c.client.DoJSON(ctx, "GET", path, nil, &wrapper); err != nil { + return nil, fmt.Errorf("配置拉取失败: %w", err) + } + return &wrapper.Data, nil +} diff --git a/backend/Dockerfile b/backend/Dockerfile index 2648659..d571fb5 100644 --- a/backend/Dockerfile +++ b/backend/Dockerfile @@ -67,12 +67,12 @@ RUN --mount=type=cache,target=/go/pkg/mod \ # 复制 agent 源码 COPY agent/ ./ -# 交叉编译三个平台的 agent 二进制（注入版本号） +# 交叉编译主流服务器平台 agent 二进制（Linux amd64 / Linux arm64 / Windows amd64，注入版本号） RUN --mount=type=cache,target=/go/pkg/mod \ --mount=type=cache,target=/root/.cache/go-build \ CGO_ENABLED=0 GOOS=linux GOARCH=amd64 go build -ldflags="-s -w -X main.version=${VERSION} -X main.buildTime=${BUILD_TIME}" -o bin/aiops-agent-linux-amd64 . && \ CGO_ENABLED=0 GOOS=linux GOARCH=arm64 go build -ldflags="-s -w -X main.version=${VERSION} -X main.buildTime=${BUILD_TIME}" -o bin/aiops-agent-linux-arm64 . && \ - CGO_ENABLED=0 GOOS=darwin GOARCH=arm64 go build -ldflags="-s -w -X main.version=${VERSION} -X main.buildTime=${BUILD_TIME}" -o bin/aiops-agent-darwin-arm64 . + CGO_ENABLED=0 GOOS=windows GOARCH=amd64 go build -ldflags="-s -w -X main.version=${VERSION} -X main.buildTime=${BUILD_TIME}" -o bin/aiops-agent-windows-amd64.exe . # ==================== Stage 3: Runtime ==================== # 固定版本避免 :latest 每次变动导致全量 cache 失效 // Logan.Li @@ -106,8 +106,10 @@ COPY --from=builder /app/aiops . # 创建运行时目录（agent 二进制 + 安装脚本） RUN mkdir -p /app/agent/bin /app/agent/deploy -# 复制 agent 二进制（U4: 仅 linux-amd64 内置, arm64/darwin 通过下载 API 提供）// Logan.Li +# 复制 agent 二进制（Linux amd64 内置，其余平台通过下载 API 提供）// Logan.Li COPY --from=agent-builder /agent/bin/aiops-agent-linux-amd64 /app/agent/bin/ +COPY --from=agent-builder /agent/bin/aiops-agent-linux-arm64 /app/agent/bin/ +COPY --from=agent-builder /agent/bin/aiops-agent-windows-amd64.exe /app/agent/bin/ # 复制 agent 安装脚本和 systemd service 模板 COPY agent/deploy/install.sh /app/agent/deploy/ diff --git a/backend/config/config.go b/backend/config/config.go index 8882606..638b2d8 100644 --- a/backend/config/config.go +++ b/backend/config/config.go @@ -97,7 +97,9 @@ func Load() *Config { viper.SetDefault("database.max_idle", 10) viper.SetDefault("jwt.expire_hour", 24) viper.SetDefault("agent.enforce_auth", false) - viper.SetDefault("retention.agent_reports_days", 7) + // 原始 Agent 上报仅作短期排障用；长期趋势由 monitor_metric_aggregates 承担， + // 故原始表保留期收敛到 1 天，避免高频表无限膨胀。 // Logan.Li + viper.SetDefault("retention.agent_reports_days", 1) viper.SetDefault("retention.metric_history_days", 30) viper.SetDefault("retention.metric_aggregate_days", 365) viper.SetDefault("retention.snmp_samples_days", 30) diff --git a/backend/controller/agent_config.go b/backend/controller/agent_config.go new file mode 100644 index 0000000..e459935 --- /dev/null +++ b/backend/controller/agent_config.go @@ -0,0 +1,198 @@ +package controller + +import ( + "crypto/sha256" + "encoding/hex" + "encoding/json" + "fmt" + "net/http" + "strconv" + "time" + + "ezops/middleware" + "ezops/model" + + "github.com/gin-gonic/gin" +) + +// Agent 采集策略的兜底默认值（表内无记录时使用）。 // Logan.Li +const ( + defaultReportInterval = 10 + defaultConfigPollInterval = 30 + // intervalFloor 与 Agent 侧校验保持一致：过密会打爆平台。 // Logan.Li + intervalFloor = 5 +) + +// GetAgentConfig 合并下发 Agent 配置。 +// +// 原先 Agent 每 5 秒要分别请求 tasks/poll、health-probes/config、port-policy 三个接口， +// 配置类请求量是上报本身的 6 倍；合并为单次响应后请求量降到原来的 1/3。 // Logan.Li +// +// GET /api/v1/agent/config +func (h *AgentHandler) GetAgentConfig(c *gin.Context) { + hostname := c.Query("hostname") + if hostname == "" { + c.JSON(http.StatusBadRequest, model.Response{Code: 400, Message: "hostname 不能为空"}) + return + } + if !middleware.RequireAgentHostname(c, hostname) { + return + } + + setting := h.resolveAgentSetting(hostname) + + // 任务：拉取 pending 并标记 running（与 /tasks/poll 行为一致）。 // Logan.Li + tasks := h.pickPendingTasks(hostname) + + // 探针配置。 // Logan.Li + probes := []model.HealthProbe{} + h.db.Where("hostname = ? AND enabled = ?", hostname, true).Order("id ASC").Find(&probes) + + // 端口策略。 // Logan.Li + policy := h.resolvePortPolicy(hostname) + + dto := model.AgentConfigDTO{ + ReportInterval: setting.ReportInterval, + ConfigPollInterval: setting.ConfigPollInterval, + Tasks: tasks, + HealthProbes: probes, + PortPolicy: policy, + } + dto.ConfigVersion = agentConfigVersion(dto) + + c.JSON(http.StatusOK, model.Response{Code: 200, Message: "success", Data: dto}) +} + +// SaveAgentSetting 平台侧设置采集策略：hostname 为空串表示全局默认。 // Logan.Li +// PUT /api/v1/agent/settings +func (h *AgentHandler) SaveAgentSetting(c *gin.Context) { + var req struct { + Hostname string `json:"hostname"` + ReportInterval int `json:"report_interval"` + ConfigPollInterval int `json:"config_poll_interval"` + } + if err := c.ShouldBindJSON(&req); err != nil { + c.JSON(http.StatusBadRequest, model.Response{Code: 400, Message: "请求格式错误: " + err.Error()}) + return + } + if req.ReportInterval < intervalFloor { + c.JSON(http.StatusBadRequest, model.Response{ + Code: 400, Message: fmt.Sprintf("上报间隔不能小于 %d 秒", intervalFloor)}) + return + } + if req.ConfigPollInterval < intervalFloor { + c.JSON(http.StatusBadRequest, model.Response{ + Code: 400, Message: fmt.Sprintf("配置轮询间隔不能小于 %d 秒", intervalFloor)}) + return + } + + setting := model.AgentSetting{Hostname: req.Hostname} + h.db.Where("hostname = ?", req.Hostname).Assign(map[string]interface{}{ + "report_interval": req.ReportInterval, + "config_poll_interval": req.ConfigPollInterval, + "updated_by": middleware.GetUsername(c), + }).FirstOrCreate(&setting) + + c.JSON(http.StatusOK, model.Response{Code: 200, Message: "已保存", Data: setting}) +} + +// ListAgentSettings 列出全局默认与所有按主机覆盖项。 // Logan.Li +// GET /api/v1/agent/settings +func (h *AgentHandler) ListAgentSettings(c *gin.Context) { + var items []model.AgentSetting + h.db.Order("hostname ASC").Find(&items) + if items == nil { + items = []model.AgentSetting{} + } + c.JSON(http.StatusOK, model.Response{Code: 200, Message: "success", Data: gin.H{ + "defaults": gin.H{ + "report_interval": defaultReportInterval, + "config_poll_interval": defaultConfigPollInterval, + }, + "items": items, + }}) +} + +// resolveAgentSetting 解析生效的采集策略：按主机覆盖 > 全局默认 > 内置默认。 // Logan.Li +func (h *AgentHandler) resolveAgentSetting(hostname string) model.AgentSetting { + effective := model.AgentSetting{ + Hostname: hostname, + ReportInterval: defaultReportInterval, + ConfigPollInterval: defaultConfigPollInterval, + } + + // 全局默认。 // Logan.Li + var global model.AgentSetting + if err := h.db.Where("hostname = ?", "").Take(&global).Error; err == nil { + applyAgentSetting(&effective, global) + } + // 按主机覆盖。 // Logan.Li + var perHost model.AgentSetting + if err := h.db.Where("hostname = ?", hostname).Take(&perHost).Error; err == nil { + applyAgentSetting(&effective, perHost) + } + return effective +} + +// applyAgentSetting 仅在来源值合法时覆盖，避免脏数据把间隔写成 0。 // Logan.Li +func applyAgentSetting(dst *model.AgentSetting, src model.AgentSetting) { + if src.ReportInterval >= intervalFloor { + dst.ReportInterval = src.ReportInterval + } + if src.ConfigPollInterval >= intervalFloor { + dst.ConfigPollInterval = src.ConfigPollInterval + } +} + +// pickPendingTasks 取出该主机待执行任务并标记 running，最多 5 个。 // Logan.Li +func (h *AgentHandler) pickPendingTasks(hostname string) []model.AgentTask { + tasks := []model.AgentTask{} + h.db.Where("hostname = ? AND status = ?", hostname, "pending"). + Order("created_at ASC").Limit(5).Find(&tasks) + if len(tasks) == 0 { + return tasks + } + now := time.Now() + for _, t := range tasks { + // 条件更新防止多 Agent 重复认领同一任务。 // Logan.Li + h.db.Model(&model.AgentTask{}). + Where("id = ? AND status = ?", t.ID, "pending"). + Updates(map[string]interface{}{"status": "running", "picked_at": now}) + } + return tasks +} + +// resolvePortPolicy 读取主机端口策略，无记录时返回空白名单。 // Logan.Li +func (h *AgentHandler) resolvePortPolicy(hostname string) *model.AgentPortPolicyDTO { + empty := &model.AgentPortPolicyDTO{Hostname: hostname, PortFilterMode: "whitelist", PortWhitelist: []int{}} + var policy model.AgentPortPolicy + if err := h.db.Where("hostname = ?", hostname).First(&policy).Error; err != nil { + return empty + } + whitelist := []int{} + _ = json.Unmarshal([]byte(policy.PortWhitelist), &whitelist) + return &model.AgentPortPolicyDTO{ + Hostname: policy.Hostname, + PortFilterMode: policy.PortFilterMode, + PortWhitelist: whitelist, + } +} + +// agentConfigVersion 对配置内容取指纹，Agent 可据此跳过未变更配置的处理。 // Logan.Li +func agentConfigVersion(dto model.AgentConfigDTO) string { + payload := struct { + ReportInterval int + ConfigPollInterval int + Tasks []model.AgentTask + HealthProbes []model.HealthProbe + PortPolicy *model.AgentPortPolicyDTO + }{ + dto.ReportInterval, dto.ConfigPollInterval, dto.Tasks, dto.HealthProbes, dto.PortPolicy, + } + raw, err := json.Marshal(payload) + if err != nil { + return strconv.FormatInt(time.Now().Unix(), 10) + } + sum := sha256.Sum256(raw) + return hex.EncodeToString(sum[:8]) +} diff --git a/backend/controller/agent_install.go b/backend/controller/agent_install.go index bb330d7..1152682 100644 --- a/backend/controller/agent_install.go +++ b/backend/controller/agent_install.go @@ -695,9 +695,23 @@ func (h *AgentInstallHandler) BatchUpgradeAgent(c *gin.Context) { binaryFile = "aiops-agent-linux-arm64" } - // 下载新二进制 + 获取版本号 + 重启服务 + // 下载新二进制 + 校验 + 重启服务。 + // 必须 fail-fast：旧版脚本用 `;` 串联并把 UPGRADE_OK 无条件 echo， + // 下载失败时同样报"成功"，导致升级结果完全不可信。 // Logan.Li upgradeScript := fmt.Sprintf( - `curl -fsSL -o /tmp/aiops-agent %s && chmod +x /tmp/aiops-agent && systemctl stop aiops-agent 2>/dev/null; mv /tmp/aiops-agent /usr/local/bin/aiops-agent && systemctl start aiops-agent; ver=$(/usr/local/bin/aiops-agent version 2>/dev/null | awk '/AIOps Agent/{print $3}'); echo "UPGRADE_OK ver=$ver"`, + `set -e; `+ + `curl -fsSL --connect-timeout 10 -o /tmp/aiops-agent.new %s || { echo "UPGRADE_FAIL download"; exit 1; }; `+ + `size=$(wc -c < /tmp/aiops-agent.new); `+ + `if [ "$size" -lt 1000000 ]; then echo "UPGRADE_FAIL size=$size"; exit 1; fi; `+ + `chmod +x /tmp/aiops-agent.new; `+ + `/tmp/aiops-agent.new version >/dev/null 2>&1 || { echo "UPGRADE_FAIL not-executable"; exit 1; }; `+ + `systemctl stop aiops-agent 2>/dev/null || true; `+ + `mv /tmp/aiops-agent.new /usr/local/bin/aiops-agent; `+ + `systemctl start aiops-agent; `+ + `sleep 1; `+ + `ver=$(/usr/local/bin/aiops-agent version 2>/dev/null | awk '/AIOps Agent/{print $3}'); `+ + `if [ -z "$ver" ]; then echo "UPGRADE_FAIL version-empty"; exit 1; fi; `+ + `echo "UPGRADE_OK ver=$ver"`, shellQuote(serverURL+"/api/agent/bin/"+binaryFile), ) diff --git a/backend/main.go b/backend/main.go index aeec060..cb246c7 100644 --- a/backend/main.go +++ b/backend/main.go @@ -23,6 +23,7 @@ import ( "ezops/pkg/cronscheduler" "ezops/pkg/database" "ezops/pkg/logger" + "ezops/pkg/metricagg" ezredis "ezops/pkg/redis" "ezops/pkg/retention" "ezops/pkg/systask" @@ -224,6 +225,10 @@ func main() { }) defer cronSched.Stop() + // 指标分层滚动聚合：raw → 1m/5m → 1h → 6h/1d，全部后台定时执行， + // 取代原先依赖前端手动触发 POST /agent/aggregate 的方式。 // Logan.Li + metricagg.Start(context.Background(), db, cache, log) + // 每日分批清理 Agent、指标、SNMP、拨测与旧日志，执行结果写入系统任务状态。 // Logan.Li go func() { ticker := time.NewTicker(24 * time.Hour) @@ -470,7 +475,7 @@ func loadConfig() *config.Config { viper.SetDefault("jwt.secret", "") // 生产环境必须通过配置或 JWT_SECRET 环境变量设置 viper.SetDefault("jwt.expire_hour", 24) viper.SetDefault("agent.enforce_auth", false) - viper.SetDefault("retention.agent_reports_days", 7) + viper.SetDefault("retention.agent_reports_days", 1) viper.SetDefault("retention.metric_history_days", 30) viper.SetDefault("retention.metric_aggregate_days", 365) viper.SetDefault("retention.snmp_samples_days", 30) diff --git a/backend/migrations/0004_metric_aggregate_unique.mysql.sql b/backend/migrations/0004_metric_aggregate_unique.mysql.sql new file mode 100644 index 0000000..d240819 --- /dev/null +++ b/backend/migrations/0004_metric_aggregate_unique.mysql.sql @@ -0,0 +1,12 @@ +-- 指标聚合表唯一约束：GORM 的 ON CONFLICT upsert 需要唯一索引才能真正生效， +-- 否则每次聚合都会插入重复行（原 idx_agg 为非唯一索引）。 // Logan.Li +-- 先按 (hostname, metric, interval, bucket_time) 去重，仅保留最大 id。 // Logan.Li +DELETE a FROM monitor_metric_aggregates a +JOIN monitor_metric_aggregates b + ON a.hostname = b.hostname + AND a.metric = b.metric + AND a.`interval` = b.`interval` + AND a.bucket_time = b.bucket_time + AND a.id < b.id; + +CREATE UNIQUE INDEX uniq_metric_agg ON monitor_metric_aggregates (hostname, metric, `interval`, bucket_time); diff --git a/backend/migrations/0004_metric_aggregate_unique.postgres.sql b/backend/migrations/0004_metric_aggregate_unique.postgres.sql new file mode 100644 index 0000000..f0e437c --- /dev/null +++ b/backend/migrations/0004_metric_aggregate_unique.postgres.sql @@ -0,0 +1,10 @@ +-- 指标聚合表唯一约束：GORM 的 ON CONFLICT upsert 需要唯一索引才能真正生效。 // Logan.Li +DELETE FROM monitor_metric_aggregates a +USING monitor_metric_aggregates b +WHERE a.hostname = b.hostname + AND a.metric = b.metric + AND a.interval = b.interval + AND a.bucket_time = b.bucket_time + AND a.id < b.id; + +CREATE UNIQUE INDEX IF NOT EXISTS uniq_metric_agg ON monitor_metric_aggregates (hostname, metric, interval, bucket_time); diff --git a/backend/migrations/0004_metric_aggregate_unique.sqlite.sql b/backend/migrations/0004_metric_aggregate_unique.sqlite.sql new file mode 100644 index 0000000..34fbcd5 --- /dev/null +++ b/backend/migrations/0004_metric_aggregate_unique.sqlite.sql @@ -0,0 +1,8 @@ +-- 指标聚合表唯一约束：GORM 的 ON CONFLICT upsert 需要唯一索引才能真正生效。 // Logan.Li +DELETE FROM monitor_metric_aggregates + WHERE id NOT IN ( + SELECT MAX(id) FROM monitor_metric_aggregates + GROUP BY hostname, metric, "interval", bucket_time + ); + +CREATE UNIQUE INDEX IF NOT EXISTS uniq_metric_agg ON monitor_metric_aggregates (hostname, metric, "interval", bucket_time); diff --git a/backend/migrations/0005_agent_settings.mysql.sql b/backend/migrations/0005_agent_settings.mysql.sql new file mode 100644 index 0000000..8e750f9 --- /dev/null +++ b/backend/migrations/0005_agent_settings.mysql.sql @@ -0,0 +1,12 @@ +-- Agent 采集策略：hostname 为空串代表全局默认，非空为按主机覆盖。 +-- 平台可通过此表下发 report_interval / config_poll_interval，Agent 热更新无需重装。 // Logan.Li +CREATE TABLE IF NOT EXISTS monitor_agent_settings ( + id BIGINT UNSIGNED NOT NULL AUTO_INCREMENT PRIMARY KEY, + hostname VARCHAR(255) NOT NULL DEFAULT '', + report_interval INT NOT NULL DEFAULT 10, + config_poll_interval INT NOT NULL DEFAULT 30, + updated_by VARCHAR(128) NOT NULL DEFAULT '', + created_at DATETIME NOT NULL, + updated_at DATETIME NOT NULL, + UNIQUE KEY uniq_agent_settings_hostname (hostname) +) ENGINE=InnoDB DEFAULT CHARSET=utf8mb4; diff --git a/backend/migrations/0005_agent_settings.postgres.sql b/backend/migrations/0005_agent_settings.postgres.sql new file mode 100644 index 0000000..4634179 --- /dev/null +++ b/backend/migrations/0005_agent_settings.postgres.sql @@ -0,0 +1,11 @@ +-- Agent 采集策略：hostname 为空串代表全局默认，非空为按主机覆盖。 // Logan.Li +CREATE TABLE IF NOT EXISTS monitor_agent_settings ( + id BIGSERIAL PRIMARY KEY, + hostname VARCHAR(255) NOT NULL DEFAULT '', + report_interval INT NOT NULL DEFAULT 10, + config_poll_interval INT NOT NULL DEFAULT 30, + updated_by VARCHAR(128) NOT NULL DEFAULT '', + created_at TIMESTAMP NOT NULL, + updated_at TIMESTAMP NOT NULL, + CONSTRAINT uniq_agent_settings_hostname UNIQUE (hostname) +); diff --git a/backend/migrations/0005_agent_settings.sqlite.sql b/backend/migrations/0005_agent_settings.sqlite.sql new file mode 100644 index 0000000..a00bc71 --- /dev/null +++ b/backend/migrations/0005_agent_settings.sqlite.sql @@ -0,0 +1,12 @@ +-- Agent 采集策略：hostname 为空串代表全局默认，非空为按主机覆盖。 // Logan.Li +CREATE TABLE IF NOT EXISTS monitor_agent_settings ( + id INTEGER PRIMARY KEY AUTOINCREMENT, + hostname VARCHAR(255) NOT NULL DEFAULT '', + report_interval INTEGER NOT NULL DEFAULT 10, + config_poll_interval INTEGER NOT NULL DEFAULT 30, + updated_by VARCHAR(128) NOT NULL DEFAULT '', + created_at DATETIME NOT NULL, + updated_at DATETIME NOT NULL +); + +CREATE UNIQUE INDEX IF NOT EXISTS uniq_agent_settings_hostname ON monitor_agent_settings (hostname); diff --git a/backend/migrations/migrations.go b/backend/migrations/migrations.go index 682a951..ffa201a 100644 --- a/backend/migrations/migrations.go +++ b/backend/migrations/migrations.go @@ -66,6 +66,20 @@ func All() []Migration { return executeDialectSQL(db, "0003_billing_settings") }, }, + { + Version: "0004_metric_aggregate_unique", + Description: "指标聚合表唯一约束（upsert 生效）", + Apply: func(db *gorm.DB) error { + return executeDialectSQL(db, "0004_metric_aggregate_unique") + }, + }, + { + Version: "0005_agent_settings", + Description: "Agent 采集策略（可下发频率）", + Apply: func(db *gorm.DB) error { + return executeDialectSQL(db, "0005_agent_settings") + }, + }, } } diff --git a/backend/model/agent_setting.go b/backend/model/agent_setting.go new file mode 100644 index 0000000..ca76fac --- /dev/null +++ b/backend/model/agent_setting.go @@ -0,0 +1,37 @@ +package model + +import "time" + +// AgentSetting Agent 采集策略。 +// +// Hostname 为空串表示全局默认，非空表示对单台主机的覆盖； +// 平台改这里即可下发新的采集/上报频率，Agent 侧热更新 ticker，无需重装。 // Logan.Li +type AgentSetting struct { + ID uint `gorm:"primarykey" json:"id"` + Hostname string `gorm:"size:255;uniqueIndex:uniq_agent_settings_hostname;not null;default:''" json:"hostname"` + ReportInterval int `gorm:"not null;default:10" json:"report_interval"` // 采集上报间隔(秒) + ConfigPollInterval int `gorm:"not null;default:30" json:"config_poll_interval"` // 配置轮询间隔(秒) + UpdatedBy string `gorm:"size:128;not null;default:''" json:"updated_by"` + CreatedAt time.Time `json:"created_at"` + UpdatedAt time.Time `json:"updated_at"` +} + +// TableName 固定表名，便于迁移脚本与代码保持一致。 // Logan.Li +func (AgentSetting) TableName() string { return "monitor_agent_settings" } + +// AgentConfigDTO 下发给 Agent 的合并配置。 +// +// 原先 Agent 每 5 秒要分别请求 tasks/poll、health-probes/config、port-policy +// 三个接口，请求量是上报本身的 6 倍；合并为单次响应后可显著降低平台与网络开销。 // Logan.Li +type AgentConfigDTO struct { + // ConfigVersion 为配置指纹，Agent 可比对后跳过后续相同配置的处理。 // Logan.Li + ConfigVersion string `json:"config_version"` + // ReportInterval 采集上报间隔(秒)，Agent 收到后热更新 ticker。 // Logan.Li + ReportInterval int `json:"report_interval"` + // ConfigPollInterval 配置轮询间隔(秒)。 // Logan.Li + ConfigPollInterval int `json:"config_poll_interval"` + + Tasks []AgentTask `json:"tasks"` + HealthProbes []HealthProbe `json:"health_probes"` + PortPolicy *AgentPortPolicyDTO `json:"port_policy,omitempty"` +} diff --git a/backend/model/metrics_aggregate.go b/backend/model/metrics_aggregate.go index d92c267..fd5f5b8 100644 --- a/backend/model/metrics_aggregate.go +++ b/backend/model/metrics_aggregate.go @@ -4,17 +4,20 @@ import "time" // MetricsAggregate 指标聚合数据（Redis 热数据 → MySQL 时序聚合）// Logan.Li // 替代旧名 MetricsKline，OHLC 对监控无意义，改用 max/min/avg/p95 +// +// (hostname, metric, interval, bucket_time) 必须是唯一索引： +// GORM 的 ON CONFLICT upsert 依赖它，否则每次聚合都会插入重复行。 // Logan.Li type MetricsAggregate struct { ID uint `gorm:"primarykey" json:"id"` - Hostname string `gorm:"size:255;index:idx_agg;not null" json:"hostname"` - Metric string `gorm:"size:32;index:idx_agg;not null" json:"metric"` // cpu / mem / load1 / disk / net_rx / net_tx - Interval string `gorm:"size:8;index:idx_agg;not null" json:"interval"` // 1m / 5m / 1h / 6h / 1d + Hostname string `gorm:"size:255;uniqueIndex:uniq_metric_agg,priority:1;not null" json:"hostname"` + Metric string `gorm:"size:32;uniqueIndex:uniq_metric_agg,priority:2;not null" json:"metric"` // cpu / mem / load1 / disk / net_rx / net_tx + Interval string `gorm:"size:8;uniqueIndex:uniq_metric_agg,priority:3;not null" json:"interval"` // 1m / 5m / 1h / 6h / 1d Avg float64 `gorm:"type:decimal(10,2)" json:"avg"` MaxV float64 `gorm:"column:max_v;type:decimal(10,2)" json:"max"` MinV float64 `gorm:"column:min_v;type:decimal(10,2)" json:"min"` P95 float64 `gorm:"type:decimal(10,2)" json:"p95"` Count int `json:"count"` - BucketTime time.Time `gorm:"index:idx_agg;not null" json:"bucket_time"` + BucketTime time.Time `gorm:"uniqueIndex:uniq_metric_agg,priority:4;not null" json:"bucket_time"` CreatedAt time.Time `json:"created_at"` } diff --git a/backend/pkg/agentauth/credential.go b/backend/pkg/agentauth/credential.go index dee42ab..2707dd1 100644 --- a/backend/pkg/agentauth/credential.go +++ b/backend/pkg/agentauth/credential.go @@ -2,9 +2,11 @@ package agentauth import ( "crypto/rand" + "crypto/sha256" "encoding/hex" "fmt" "strings" + "sync" "time" "ezops/model" @@ -13,6 +15,120 @@ import ( "gorm.io/gorm" ) +// ───────────────────────────────────────────────────────────── +// Token 校验缓存 +// +// bcrypt 被设计成故意慢（本项目实测单次 Cost=10 约 72ms CPU）。Agent 每 10 秒 +// 轮询 3 个接口，每个接口都要校验一次 Token，30 台主机即 9 req/s×72ms ≈ 0.65 核， +// 纯粹烧在重复校验同一条不可变凭证上。 +// +// 由于 Token 在轮换前不可变（Issue 才会改写 token_hash），校验结果可以安全缓存： +// - 缓存键：token 的 SHA-256（不落明文，避免内存里出现可直接使用的凭证） +// - 有效期：默认 5 分钟，过期后回落一次 bcrypt 并续期 +// - 轮换：Issue 时全量清空，保证「旧 Token 立即失效」语义不变 +// +// last_used_at 的写入同样按前缀节流（默认 60s 一次），避免每请求一次 UPDATE。 +// // Logan.Li +type cachedCredential struct { + credential *model.AgentCredential + expiresAt time.Time + // lastWrite 记录最近一次把 last_used_at 落库的时间，用于写入节流。 // Logan.Li + lastWrite time.Time +} + +const ( + defaultCacheTTL = 5 * time.Minute + defaultWriteThrottle = 60 * time.Second + // cacheMaxEntries 上限，防止异常 Token 洪水把内存撑爆。 // Logan.Li + cacheMaxEntries = 4096 +) + +var ( + credCacheMu sync.Mutex + credCache = make(map[string]*cachedCredential) + // cacheTTL / writeThrottle 允许测试覆盖。 // Logan.Li + cacheTTL = defaultCacheTTL + writeThrottle = defaultWriteThrottle +) + +// cacheKey 只缓存 Token 的哈希，内存中不保留明文 Token。 // Logan.Li +func cacheKey(token string) string { + sum := sha256.Sum256([]byte(token)) + return hex.EncodeToString(sum[:]) +} + +// lookupCache 返回命中的凭证副本；命中时同时按节流更新 last_used_at。 // Logan.Li +func lookupCache(db *gorm.DB, token string) (*model.AgentCredential, bool) { + key := cacheKey(token) + credCacheMu.Lock() + defer credCacheMu.Unlock() + + entry, ok := credCache[key] + if !ok || time.Now().After(entry.expiresAt) { + if ok { + delete(credCache, key) + } + return nil, false + } + if entry.credential.ExpiresAt != nil && entry.credential.ExpiresAt.Before(time.Now()) { + delete(credCache, key) + return nil, false + } + + // 节流写入 last_used_at：仅当距上次落库超过阈值才真正 UPDATE。 // Logan.Li + now := time.Now() + if now.Sub(entry.lastWrite) >= writeThrottle { + entry.lastWrite = now + prefix := entry.credential.TokenPrefix + go func() { + _ = db.Model(&model.AgentCredential{}). + Where("token_prefix = ?", prefix). + Update("last_used_at", now).Error + }() + } + + cp := *entry.credential + cp.LastUsedAt = &now + return &cp, true +} + +// storeCache 写入缓存，必要时按容量上限淘汰过期项。 // Logan.Li +func storeCache(token string, credential *model.AgentCredential) { + key := cacheKey(token) + credCacheMu.Lock() + defer credCacheMu.Unlock() + + if len(credCache) >= cacheMaxEntries { + now := time.Now() + for k, v := range credCache { + if now.After(v.expiresAt) { + delete(credCache, k) + } + } + // 仍超上限则整体清空，宁可多算几次 bcrypt 也不无限占内存。 // Logan.Li + if len(credCache) >= cacheMaxEntries { + credCache = make(map[string]*cachedCredential) + } + } + + cp := *credential + credCache[key] = &cachedCredential{ + credential: &cp, + expiresAt: time.Now().Add(cacheTTL), + lastWrite: time.Now(), + } +} + +// invalidateCache 清空全部缓存。Token 轮换/撤销后调用，保证旧 Token 立即失效。 // Logan.Li +func invalidateCache() { + credCacheMu.Lock() + credCache = make(map[string]*cachedCredential) + credCacheMu.Unlock() +} + +// Invalidate 提供给凭证变更入口（轮换/撤销/删除）显式失效缓存。 // Logan.Li +func Invalidate() { invalidateCache() } + // Issue 为指定主机轮换 Agent Token，旧 Token 会立即失效。 // Logan.Li func Issue(db *gorm.DB, hostname, issuedBy string) (string, error) { hostname = strings.TrimSpace(hostname) @@ -47,11 +163,20 @@ func Issue(db *gorm.DB, hostname, issuedBy string) (string, error) { if err != nil { return "", err } + // 轮换后必须让缓存失效，否则被缓存的旧 Token 还能继续通过校验。 // Logan.Li + invalidateCache() return token, nil } // Authenticate 校验 Token 并返回绑定的主机名。 // Logan.Li +// +// 命中缓存时跳过 bcrypt 与凭证查询，只做一次可选节流的 last_used_at 更新； +// 未命中时才回落完整校验路径。 // Logan.Li func Authenticate(db *gorm.DB, token string) (*model.AgentCredential, error) { + if credential, ok := lookupCache(db, token); ok { + return credential, nil + } + parts := strings.SplitN(token, ".", 2) if len(parts) != 2 || !strings.HasPrefix(parts[0], "agt_") || parts[1] == "" { return nil, fmt.Errorf("invalid agent token") @@ -69,6 +194,9 @@ func Authenticate(db *gorm.DB, token string) (*model.AgentCredential, error) { now := time.Now() _ = db.Model(&credential).Update("last_used_at", now).Error credential.LastUsedAt = &now + + // 校验通过后写入缓存，后续请求直接命中。 // Logan.Li + storeCache(token, &credential) return &credential, nil } diff --git a/backend/pkg/metricagg/aggregator.go b/backend/pkg/metricagg/aggregator.go new file mode 100644 index 0000000..d8a4211 --- /dev/null +++ b/backend/pkg/metricagg/aggregator.go @@ -0,0 +1,311 @@ +// Package metricagg 指标分层滚动聚合。 +// +// 目标：把高频原始采样（Redis ZSET）逐级降采样为多分辨率聚合行，落库供趋势查询。 +// +// 层级（逐级降采样，每级只读上一级，避免每次从原始数据重算）： +// +// raw(10s, Redis) --1min--> 1m --1h--> 1h --6h/24h--> 6h / 1d +// +// 调度（全部后台，替代原先依赖前端手动触发 POST /agent/aggregate）： +// - fine 每 1 分钟：raw → 1m / 5m +// - mid 每 15 分钟：1m(DB) → 1h +// - coarse 每 1 小时：1h(DB) → 6h / 1d +// +// 幂等：依赖 monitor_metric_aggregates 的 uniq_metric_agg 唯一索引 + ON CONFLICT。 // Logan.Li +package metricagg + +import ( + "context" + "math" + "sort" + "strconv" + "time" + + "ezops/model" + ezredis "ezops/pkg/redis" + + "go.uber.org/zap" + "gorm.io/gorm" + "gorm.io/gorm/clause" +) + +// metrics 需要聚合的指标清单（与 Agent 写入 Redis 的 key 保持一致）。 // Logan.Li +var metrics = []string{"cpu", "mem", "load1"} + +const ( + // rawKeyPrefix 与 controller.writeHotMetric 写入的 key 前缀保持一致。 // Logan.Li + rawKeyPrefix = "metrics:" + + fineEvery = time.Minute + midEvery = 15 * time.Minute + coarseEvery = time.Hour + + // 各级回看窗口：只需覆盖该级「刚闭合的桶 + 边界抖动」。 // Logan.Li + fineLookback = 3 * time.Hour + midLookback = 6 * time.Hour + coarseLookback = 8 * 24 * time.Hour +) + +// fineIntervals / midIntervals / coarseIntervals 各级负责的聚合周期。 // Logan.Li +var ( + fineIntervals = []interval{{"1m", time.Minute}, {"5m", 5 * time.Minute}} + midIntervals = []interval{{"1h", time.Hour}} + coarseIntervals = []interval{{"6h", 6 * time.Hour}, {"1d", 24 * time.Hour}} +) + +type interval struct { + name string + d time.Duration +} + +// Start 启动后台聚合调度，ctx 取消后退出。 // Logan.Li +func Start(ctx context.Context, db *gorm.DB, cache ezredis.Cache, log *zap.Logger) { + // 无 ZSet 能力（redka/local 模式）时不做 Redis 侧聚合。 // Logan.Li + if cache == nil || !cache.Capabilities().SortedSet { + log.Info("指标聚合未启用：当前缓存后端不支持有序集合") + return + } + + go run(ctx, db, cache, log, fineEvery, "fine", func(c context.Context) { + rollupFromRaw(c, db, cache, log, fineIntervals, fineLookback) + }) + go run(ctx, db, cache, log, midEvery, "mid", func(c context.Context) { + rollupFromDB(db, log, "1m", midIntervals, midLookback) + }) + go run(ctx, db, cache, log, coarseEvery, "coarse", func(c context.Context) { + rollupFromDB(db, log, "1h", coarseIntervals, coarseLookback) + }) +} + +// run 周期性执行任务，起跑前先等一个周期，避免与启动期初始化抢资源。 // Logan.Li +func run(ctx context.Context, _ *gorm.DB, _ ezredis.Cache, log *zap.Logger, every time.Duration, name string, fn func(context.Context)) { + ticker := time.NewTicker(every) + defer ticker.Stop() + for { + select { + case <-ctx.Done(): + return + case <-ticker.C: + start := time.Now() + fn(ctx) + log.Debug("指标聚合完成", zap.String("level", name), zap.Duration("cost", time.Since(start))) + } + } +} + +// rollupFromRaw 从 Redis 原始采样聚合出细粒度桶。 // Logan.Li +func rollupFromRaw(ctx context.Context, db *gorm.DB, cache ezredis.Cache, log *zap.Logger, ivs []interval, lookback time.Duration) { + hosts := activeHosts(db) + if len(hosts) == 0 { + return + } + now := time.Now() + since := now.Add(-lookback) + + for _, host := range hosts { + for _, metric := range metrics { + key := rawKeyPrefix + metric + ":" + host + members, err := cache.ZRangeByScore(ctx, key, + formatUnix(since), formatUnix(now.Add(time.Minute))) + if err != nil || len(members) == 0 { + continue + } + for _, iv := range ivs { + buckets := bucketizeMembers(members, iv.d) + if err := upsertBuckets(db, host, metric, iv.name, buckets); err != nil { + log.Warn("写入聚合桶失败", zap.String("host", host), zap.String("metric", metric), zap.Error(err)) + } + } + } + } +} + +// rollupFromDB 从上一级聚合行再聚合出粗粒度桶（逐级降采样）。 // Logan.Li +func rollupFromDB(db *gorm.DB, log *zap.Logger, sourceInterval string, ivs []interval, lookback time.Duration) { + hosts := activeHosts(db) + if len(hosts) == 0 { + return + } + since := time.Now().Add(-lookback) + + for _, host := range hosts { + for _, metric := range metrics { + var rows []model.MetricsAggregate + if err := db.Where("hostname = ? AND metric = ? AND `interval` = ? AND bucket_time >= ?", + host, metric, sourceInterval, since). + Order("bucket_time ASC").Find(&rows).Error; err != nil || len(rows) == 0 { + continue + } + samples := make([]weightedSample, 0, len(rows)) + for _, r := range rows { + samples = append(samples, weightedSample{ts: r.BucketTime, avg: r.Avg, count: r.Count}) + } + for _, iv := range ivs { + buckets := bucketizeSamples(samples, iv.d) + if err := upsertBuckets(db, host, metric, iv.name, buckets); err != nil { + log.Warn("写入降采样桶失败", zap.String("host", host), zap.String("metric", metric), zap.Error(err)) + } + } + } + } +} + +// activeHosts 只聚合近期在线的 Agent 主机，避免对已下线主机做无用扫描。 // Logan.Li +func activeHosts(db *gorm.DB) []string { + var hosts []string + cutoff := time.Now().Add(-15 * time.Minute) + db.Model(&model.Server{}). + Where("agent_status = ? AND last_check_at >= ?", "online", cutoff). + Distinct().Pluck("hostname", &hosts) + return hosts +} + +type bucketStats struct { + avg, maxV, minV, p95 float64 + count int +} + +type weightedSample struct { + ts time.Time + avg float64 + count int +} + +// bucketizeMembers 把原始采样按周期分桶。 // Logan.Li +func bucketizeMembers(members []ezredis.SortedSetMember, d time.Duration) map[time.Time][]float64 { + sec := int64(d.Seconds()) + out := make(map[time.Time][]float64) + for _, m := range members { + ts := int64(m.Score) + start := (ts / sec) * sec + out[time.Unix(start, 0).UTC()] = append(out[time.Unix(start, 0).UTC()], parseValue(m.Member)) + } + return out +} + +// bucketizeSamples 把上一级聚合行按周期再聚合（用 count 加权，避免小样本被放大）。 // Logan.Li +func bucketizeSamples(samples []weightedSample, d time.Duration) map[time.Time][]float64 { + sec := int64(d.Seconds()) + out := make(map[time.Time][]float64) + for _, s := range samples { + ts := s.ts.Unix() + start := (ts / sec) * sec + key := time.Unix(start, 0).UTC() + // 用 avg 重复 count 次近似还原样本分布，保证 max/min/p95 有意义。 // Logan.Li + n := s.count + if n < 1 { + n = 1 + } + if n > 1000 { + n = 1000 + } + for i := 0; i < n; i++ { + out[key] = append(out[key], s.avg) + } + } + return out +} + +// upsertBuckets 幂等写入聚合行（依赖 uniq_metric_agg 唯一索引）。 // Logan.Li +func upsertBuckets(db *gorm.DB, host, metric, intervalName string, buckets map[time.Time][]float64) error { + for bucketTime, values := range buckets { + if len(values) == 0 { + continue + } + st := calcStats(values) + row := model.MetricsAggregate{ + Hostname: host, + Metric: metric, + Interval: intervalName, + Avg: st.avg, + MaxV: st.maxV, + MinV: st.minV, + P95: st.p95, + Count: st.count, + BucketTime: bucketTime, + } + if err := db.Clauses(clause.OnConflict{ + Columns: []clause.Column{ + {Name: "hostname"}, {Name: "metric"}, {Name: "interval"}, {Name: "bucket_time"}, + }, + DoUpdates: clause.AssignmentColumns([]string{"avg", "max_v", "min_v", "p95", "count"}), + }).Create(&row).Error; err != nil { + return err + } + } + return nil +} + +// calcStats 计算 avg / max / min / p95，结果保留两位小数。 // Logan.Li +func calcStats(values []float64) bucketStats { + sorted := make([]float64, len(values)) + copy(sorted, values) + sort.Float64s(sorted) + + var sum float64 + for _, v := range sorted { + sum += v + } + avg := sum / float64(len(sorted)) + p95 := sorted[int(math.Ceil(float64(len(sorted))*0.95))-1] + + return bucketStats{ + avg: round2(avg), + maxV: round2(sorted[len(sorted)-1]), + minV: round2(sorted[0]), + p95: round2(p95), + count: len(sorted), + } +} + +func round2(v float64) float64 { return math.Round(v*100) / 100 } + +// parseValue 从 "45.2300:1234567890123456789" 中取出数值部分。 // Logan.Li +func parseValue(member string) float64 { + for i := 0; i < len(member); i++ { + if member[i] == ':' { + return parseFloat(member[:i]) + } + } + return parseFloat(member) +} + +func parseFloat(s string) float64 { + var v float64 + var frac, div float64 = 0, 1 + neg := false + i := 0 + if i < len(s) && (s[i] == '-' || s[i] == '+') { + neg = s[i] == '-' + i++ + } + for ; i < len(s); i++ { + c := s[i] + if c == '.' { + i++ + break + } + if c < '0' || c > '9' { + break + } + v = v*10 + float64(c-'0') + } + for ; i < len(s); i++ { + c := s[i] + if c < '0' || c > '9' { + break + } + frac = frac*10 + float64(c-'0') + div *= 10 + } + v += frac / div + if neg { + v = -v + } + return v +} + +func formatUnix(t time.Time) string { + // Redis ZSET score 为 Unix 秒；用整数文本作为 ZRANGEBYSCORE 的边界。 // Logan.Li + return strconv.FormatInt(t.Unix(), 10) +} diff --git a/backend/router/agent_static.go b/backend/router/agent_static.go index 5f641fb..cb83a05 100644 --- a/backend/router/agent_static.go +++ b/backend/router/agent_static.go @@ -19,7 +19,6 @@ type agentStaticPaths struct { var agentBinaryFiles = map[string]string{ "linux-amd64": "aiops-agent-linux-amd64", "linux-arm64": "aiops-agent-linux-arm64", - "darwin-arm64": "aiops-agent-darwin-arm64", "windows-amd64": "aiops-agent-windows-amd64.exe", } diff --git a/backend/router/routes.go b/backend/router/routes.go index ab511e1..5b681e8 100644 --- a/backend/router/routes.go +++ b/backend/router/routes.go @@ -886,6 +886,9 @@ func Register(r *gin.Engine, db *gorm.DB, log *zap.Logger, cfg *config.Config, c agentGroup.GET("/health-probes/config", agentHandler.GetHealthProbesForAgent) // 端口策略：Agent 拉取本机端口过滤策略。 agentGroup.GET("/port-policy", agentHandler.GetPortPolicyForAgent) + // 合并配置：一次返回任务 + 探针 + 端口策略 + 采集频率， + // 取代原先每 5 秒分别拉取上面三个接口的做法。 // Logan.Li + agentGroup.GET("/config", agentHandler.GetAgentConfig) } // Agent 数据查询 + 任务管理（需要 JWT） agentQuery := authorized.Group("/agent") @@ -918,6 +921,10 @@ func Register(r *gin.Engine, db *gorm.DB, log *zap.Logger, cfg *config.Config, c agentQuery.POST("/batch-uninstall", agentInstallHandler.BatchUninstallAgent) agentQuery.POST("/batch-upgrade", agentInstallHandler.BatchUpgradeAgent) + // 采集策略（平台下发上报/轮询频率，Agent 热更新无需重装） + agentQuery.GET("/settings", agentHandler.ListAgentSettings) + agentQuery.PUT("/settings", agentHandler.SaveAgentSetting) + // 健康探针管理（CRUD） agentQuery.GET("/health-probes", agentHandler.ListHealthProbes) agentQuery.POST("/health-probes", agentHandler.CreateHealthProbe) diff --git a/run.sh b/run.sh index 2be2ddd..6dd899c 100755 --- a/run.sh +++ b/run.sh @@ -864,8 +864,6 @@ build_agent_binaries() { GOOS=linux GOARCH=amd64 CGO_ENABLED=0 go build -ldflags="-w -s" -o bin/aiops-agent-linux-amd64 . info "编译 Agent Linux ARM64..." GOOS=linux GOARCH=arm64 CGO_ENABLED=0 go build -ldflags="-w -s" -o bin/aiops-agent-linux-arm64 . - info "编译 Agent macOS ARM64..." - GOOS=darwin GOARCH=arm64 CGO_ENABLED=0 go build -ldflags="-w -s" -o bin/aiops-agent-darwin-arm64 . info "编译 Agent Windows AMD64..." GOOS=windows GOARCH=amd64 CGO_ENABLED=0 go build -ldflags="-w -s" -o bin/aiops-agent-windows-amd64.exe . success "Agent 多平台二进制编译完成: agent/bin/"