groups: - name: ai-ticket-platform rules: - alert: AiTicketBackendDown expr: up{job="ai-ticket-backend"} == 0 for: 2m labels: severity: critical annotations: summary: AI 工单平台后端不可用 description: Prometheus 连续 2 分钟无法抓取后端指标。 - alert: AiTicketChannelDeliveryFailures expr: ai_ticket_channel_messages_failed_recent > 0 for: 5m labels: severity: warning annotations: summary: 渠道消息投递持续失败 description: 最近 15 分钟存在失败的渠道消息,请检查渠道凭证、地址和重试队列。 - alert: AiTicketChannelQueueStuck expr: ai_ticket_channel_messages_stuck > 0 for: 5m labels: severity: critical annotations: summary: 渠道消息队列出现积压 description: 存在排队超过 10 分钟的出站消息,请检查 Celery Worker 和渠道网络。 - alert: AiTicketAgentRunFailures expr: ai_ticket_agent_runs_failed_recent > 0 for: 5m labels: severity: warning annotations: summary: Agent 运行持续失败 description: 最近 15 分钟存在失败的 Agent 运行,请检查模型、工具和运行日志。 - alert: AiTicketToolCallFailures expr: ai_ticket_tool_calls_failed_recent > 0 for: 5m labels: severity: warning annotations: summary: Tool/MCP 调用持续失败 description: 最近 15 分钟存在失败的工具调用,请检查授权、MCP 健康和审计记录。 - alert: AiTicketWorkItemBacklog expr: ai_ticket_work_items_active > 200 for: 30m labels: severity: warning annotations: summary: 活动任务积压 description: 活动任务连续 30 分钟超过 200 个,请检查人工处理容量和自动任务异常。