<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Distributed-Systems on Osman Goni Nahid</title><link>https://osmangoninahid.github.io/tags/distributed-systems/</link><description>Recent content in Distributed-Systems on Osman Goni Nahid</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Tue, 29 Sep 2026 00:00:00 +0400</lastBuildDate><atom:link href="https://osmangoninahid.github.io/tags/distributed-systems/index.xml" rel="self" type="application/rss+xml"/><item><title>Stuck in Undeploying: when your status and your cluster disagree</title><link>https://osmangoninahid.github.io/stuck-undeploying-two-ledgers/</link><pubDate>Tue, 29 Sep 2026 00:00:00 +0400</pubDate><guid>https://osmangoninahid.github.io/stuck-undeploying-two-ledgers/</guid><description>&lt;p&gt;&amp;ldquo;Stuck in Undeploying.&amp;rdquo; &amp;ldquo;Stuck in Pending.&amp;rdquo; &amp;ldquo;Stuck in Terminating.&amp;rdquo; Over a year, that word showed up in more than a hundred tickets on a platform that deploys models onto GPU clusters. The states were different; the shape of the problem was the same every time.&lt;/p&gt;&#10;&lt;figure&gt;&lt;img src="https://osmangoninahid.github.io/images/stuck-two-ledgers.gif" alt="Two ledgers side by side: the platform database says Undeploying for hours while Kubernetes shows the pod gone, a PVC waiting on a finalizer and GPU memory still held; a reconcile loop brings them back in line" loading="lazy"&gt;&lt;/figure&gt;&#10;&#10;&lt;p&gt;A platform keeps its own record of what a workload is doing. Kubernetes keeps the truth. &amp;ldquo;Stuck&amp;rdquo; is the gap between them.&lt;/p&gt;</description></item></channel></rss>