Using Kubernetes for handling second screen experience of european tv show About us Thomas Hector Head of IT Operations Key Qualifications ● 12+ years’ experience with development and operations operations ● Automating everything ● Been handling Kubernetes clusters in production for 2 years Jan-Erik Revsbech CTO Key Qualifications ● 20+ year of experience in web-development & operation/hosting

the following mechanisms to handle the component launch sequence: – Init Container: a pre-handling container to process staff which need to be done before the major costainer starts – Probe:

reached 1.0 and has been released as stable/GA with Kubernetes v1.13. It is now the suggested way of handling storage volumes inside Kubernetes clusters. Status within the Kubernetes project 10 How to contribute

failedPodContr oller Detector Strategy Unhealthy node list Fast Taint Weight Adjust Recovery Manual Handling Improve Auto Human experience Improve of strategy …… 1. Collect data from metrics NPD, Trace

timeoutSeconds: 1 name: liveness 容器的源码： http.HandleFunc("/healthz", func(w http.ResponseWriter, r *http.Request) { duration := time.Now().Sub(started) if duration.Seconds() > 10 { w.WriteHeader(500) 本⽂档强调并整合了整个⽤户指南、⼊⻔⽂档和示例中引⼊的配置最佳实践。 这是⼀个“活”的⽂件。如果你想到的东⻄不在这个名单上，但可能对他⼈有⽤，请不要犹豫，提交issue或提交PR（pull request）。 General Config Tips（常规配置Tips） 定义配置时，指定最新的稳定API版本（当前为v1）。 配置⽂件应该被推送到集群之前，应存储在版本控制软件中。这样，如果需要，可以快速回滚配置。如果需要，还 ion/overview/ 19-配置最佳实践 81 Managing Compute Resources for Containers（管理容 器的计算资源） 译者按：本节中，笔者将request翻译成最⼩需求，limit翻译成最⼤限制。由于出现的次数太多，故⽽绝⼤多数地 ⽅直接不翻译了，⼤家可以当做术语来阅读。 指定 Pod 时，可选择指定每个容器需要多少CPU和内存（RAM）。当

Node2 Kube-scheduler Pod Request Load Level Request Load Level Real Load Level Real Load Level Assigned to Node2 The native K8S scheduling is based on the resource request of Pod. However, in many resource requests but low load. Dynamic-Scheduler Node1 Node2 Kube-scheduler Pod Request Load Level Request Load Level Real Load Level Real Load Level Assigned to Node1 Dynamic-scheduler Node1

⽀持容器组（Pod）的弹性伸缩，即根据容器 CPU 和内存资源占⽤情况⾃动调整容器副本数 量。 指标：可选 CPU 使⽤量和内存使⽤量，需要和设置的所需资源类型相同。 触发条件：资源使⽤率的百分⽐，超过设置的Pod request值，容器开始扩容。 最⼤副本数：该 Deployment 可扩容的副本（即容器组Pod）数量上限。 最⼩副本数：该 Deployment 可缩容的副本（即容器组Pod）数量下限。 注：若要 ⽀持容器组（Pod）的弹性伸缩，即根据容器 CPU 和内存资源占⽤情况⾃动调整容器副本数 量。 指标：可选 CPU 使⽤量和内存使⽤量，需要和设置的所需资源类型相同。 触发条件：资源使⽤率的百分⽐，超过设置的Pod request值，容器开始扩容。 最⼤副本数：该 Deployment 可扩容的副本（即容器组Pod）数量上限。 最⼩副本数：该 Deployment 可缩容的副本（即容器组Pod）数量下限。 注：若要 请求（Request）进⾏计算的。当Pod由于资源请求（Request）⽆法满⾜并进⼊等待（Pending）状态 时，节点⾃动伸缩组件会根据配置的弹性伸缩组配置信息中的资源规格以及约束配置，计算所需的节点数 ⽬，如果可以满⾜伸缩条件，则会触发伸缩组的节点加⼊。当⼀个节点在弹性伸缩组中且节点上Pod的资 源请求低于阈值时，节点⾃动伸缩组件会将节点进⾏缩容。因此资源请求（Request）的正确、合理设

js var http = require('http'); var handleRequest = function(request, response) { console.log('Received request for URL: ' + request.url); response.writeHead(200); response.end('Hello World

Ø DynamicQuota-Operator to reconcile business quota. Ø ValidatingWebhook to validate pod add request. DynamicQuota Large-scale and high-performance Autoscaling Motivation: Ø Reduce container OOM Performance. Ø Personalized configuration of HPA objects. Ø Calculate replicas based on pod resource request or limit. Ø Cooperate with CronHPA and VWA. Ø Dynamic adjust ‘minReplicas’ field based on historical

YARN Application centric: top down. Scheduling sequence: Queue -> user -> application -> container request ������������� kubernetes Based on api-server watch mechanism Everything stored in etcd YARN Based