Ask HN：将 Kubernetes 依赖项连接到应用程序遥测
3 分•作者: chipfixer•20 天前
我正在开发 Cardinal (cardinalhq.io)。Kubernetes 调试中最让我头疼的一点是，从应用程序问题到其底层 Kubernetes 根本原因需要经过的步骤太多了。
例如，如果我看到错误率飙升，并且按部署分组，我希望立即看到该部署的 Kubernetes "邻居"（k8s neighborhood），然后就能显示最近被错误更新的 ConfigMap。
同样，假设一个节点磁盘空间不足，这导致一个 StatefulSet 在尝试写入新文件时产生连锁反应，进而影响请求延迟。
诸如此类的情况不胜枚举。
为了解决这个问题，我们有一个想法：构建一个智能图表图例，它能识别 Kubernetes 标签（基于 OTEL 语义约定），然后为每个条目弹出一个“基础设施地图”（infra map）图标。点击后，您就能看到该实体的完整 Kubernetes "邻居" 以及健康状况注解。这样，从应用程序遥测数据关联到 Kubernetes 基础设施就变得更容易了。
我想了解一下这种用户体验是否合理，如果不合理，您目前是如何解决这种关联问题的？
https://cardinalhq.io/use-cases/kubernetes
查看原文
I’m building Cardinal (cardinalhq.io). One thing that drives me nuts about Kubernetes debugging is the number of steps it takes to go from an application issue to its underlying kubernetes root cause.<p>Like for example, if I am seeing an error rate spike, and I group by deployment, I want to immediately see the deployment's k8s neighborhood, which would then show me the configmap that was updated recently and incorrectly.<p>Similarly, say a node is low on disk and that is causing a ripple effect on a statefulset trying to write new files, which then affects request latency.<p>The list goes on.<p>To solve this, one idea we had was: build a smart chart legend that recognizes a k8s tag (based on OTEL semantic conventions), and then for every entry pops an "infra map" icon. Click and you see the full k8s neighborhood of that entity with health annotations. So now it's easier to go from app telemetry to k8s infra.<p>I am curious if this UX makes sense, and if not how do you solve this correlation problem today?<p>https://cardinalhq.io/use-cases/kubernetes