cl-lightning-talk-how-prometheus-ai-agent-helps-build-interactive-monitoring-cl-cong-shu-xi-prometheus-ailia-lai-dyags-zhihao-liu-quwan.pdf

《cl-lightning-talk-how-prometheus-ai-agent-helps-build-interactive-monitoring-cl-cong-shu-xi-prometheus-ailia-lai-dyags-zhihao-liu-quwan.pdf》由会员分享，可在线阅读，更多相关《cl-lightning-talk-how-prometheus-ai-agent-helps-build-interactive-monitoring-cl-cong-shu-xi-prometheus-ailia-lai-dyags-zhihao-liu-quwan.pdf（12页珍藏版）》请在三个皮匠报告上搜索。

1、Build Interactive MonitoringZhihao Liu QuwanDevops Engineer-Prometheus AI Agent Current SituationWhat are the metrics for this instance resource?这个实例资源有哪些指标？How to write PromQL for this?如何写的PromQL？For this Alert/Event,which metrics should I check?这个告警/事件，我应该去看哪些有效指标？How can I view the chart of relat

2、ed resources?如何查看关联资源的面板？Can it be viewed quickly on mobile devices?可以快捷移动端查看吗？.Can LLM solve these problems?bring a new interactive experience?LLM能解决这些问题吗？能带来新的交互体验吗？Question:ExploreText-to-PromQLRetrieval Augmented Generation(RAG)Like HumanAI AgentExploreBut LLM has its own ideas?(Hallucination)但是

3、大模型常常有它自己的想法?（幻觉幻觉）User questions are casual and ambiguous.用户的提问是随性的，模凌两可的 Hallucinations occur,resulting in low accuracy and poor relevance.出现幻觉，结果准确性低，相关性差 The Chat method is cumbersome.对答的方式很费劲 Writing PromQL is just a means,not the end goal.写PromQL只是目标不是目的ThoughtMeAI AgentDetermine the query obj

4、ective确定查询目标Query RewriteSearch Google and look through communities搜索谷歌、社区寻找MultiRetriever and Re-RankReference to write PromQL and execute query参考编写PromQL，执行查询LogsRetrieverWhen errors occur,analyze and adjust PromQL遇到错误，分析，调整PromQLReAct Check And Generated PromQLsView the results查看结果Create ChartQuery RewriteMultiRetriever and Re-RankLogsRetrieverLLM needs examples.Good CaseReAct Check And Generated PromQLsCreate ChartGeneralShow CaseThe End!