<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:content="http://purl.org/rss/1.0/modules/content/"><channel><title>GlobalRouter on SRE Engineering Practice</title><link>https://www.sre.wang/en/tags/globalrouter/</link><description>Recent content in GlobalRouter on SRE Engineering Practice</description><image><title>SRE Engineering Practice</title><url>https://www.sre.wang/share-card-square.jpg</url><link>https://www.sre.wang/share-card-square.jpg</link></image><generator>Hugo -- gohugo.io</generator><language>en</language><lastBuildDate>Tue, 08 Sep 2026 10:32:49 +0800</lastBuildDate><atom:link href="https://www.sre.wang/en/tags/globalrouter/index.xml" rel="self" type="application/rss+xml"/><item><title>Pods Stuck Pending After IP Pool Exhaustion: 5 Production Decisions for TKE Network Selection and Node Pool Governance</title><link>https://www.sre.wang/en/posts/cloud-tke-network-nodepool-governance/</link><pubDate>Tue, 08 Sep 2026 10:32:49 +0800</pubDate><guid>https://www.sre.wang/en/posts/cloud-tke-network-nodepool-governance/</guid><description>Starting from a real production incident where VPC-CNI elastic network interface IP exhaustion caused pods to enter Pending en masse, this article breaks down TKE&amp;#39;s three network model selection boundaries, VPC-CNI IP pool pre-binding tuning, node pool autoscaling silent failure troubleshooting, mixed-mode governance, and node type selection — delivering 5 actionable production governance decisions.</description></item></channel></rss>