<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Cloud-Init on Karthik B Hegde | DevOps Engineer</title><link>https://karthikhegde.in/tags/cloud-init/</link><description>Recent content in Cloud-Init on Karthik B Hegde | DevOps Engineer</description><generator>Hugo -- gohugo.io</generator><language>en</language><copyright>© 2026 Karthik B Hegde</copyright><lastBuildDate>Wed, 05 Aug 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://karthikhegde.in/tags/cloud-init/index.xml" rel="self" type="application/rss+xml"/><item><title>The Node That Wouldn't Come Back: Recovering a K3s Cluster on Oracle's Free Tier</title><link>https://karthikhegde.in/blog/k3s-node-recovery/</link><pubDate>Wed, 05 Aug 2026 00:00:00 +0000</pubDate><guid>https://karthikhegde.in/blog/k3s-node-recovery/</guid><description>&lt;p&gt;On 29 July my K3s node stopped responding. Rebuilding it should have been a ten-minute &lt;code&gt;terraform apply&lt;/code&gt;. It took five days.&lt;/p&gt;
&lt;p&gt;Not because the recovery itself was complicated, but because three assumptions I had never actually tested turned out to be wrong at the same time — and two of them failed &lt;em&gt;silently&lt;/em&gt;. The cluster came back looking healthy while quietly missing half its telemetry. This is what broke, why each assumption held right up until it didn&amp;rsquo;t, and what I changed so the same sequence can&amp;rsquo;t repeat.&lt;/p&gt;</description></item></channel></rss>