<?xml version="1.0" encoding="utf-8" standalone="yes"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
  <channel>
    <title>Reliability on Shawn Sorichetti</title>
    <link>/tags/reliability/</link>
    <description>Recent content in Reliability on Shawn Sorichetti</description>
    <generator>Hugo -- gohugo.io</generator>
    <language>en</language>
    <managingEditor>me@ssoriche.com (Shawn Sorichetti)</managingEditor>
    <webMaster>me@ssoriche.com (Shawn Sorichetti)</webMaster>
    <copyright>© 2026 Shawn Sorichetti</copyright>
    <lastBuildDate>Sun, 14 Jun 2026 10:21:43 -0400</lastBuildDate><atom:link href="/tags/reliability/index.xml" rel="self" type="application/rss+xml" />
    
    <item>
      <title>A Readiness Probe That Can&#39;t Fail Is Just Wallpaper</title>
      <link>/til/kubernetes-readiness-probe-wallpaper/</link>
      <pubDate>Mon, 08 Jun 2026 00:00:00 +0000</pubDate>
      <author>me@ssoriche.com (Shawn Sorichetti)</author>
      <guid>/til/kubernetes-readiness-probe-wallpaper/</guid>
      <description>&lt;p&gt;Writing up a production outage report today, I was making the case that workloads should fail their readiness probes when they can&amp;rsquo;t reach their dependencies — databases, caches, anything required to do useful work. The collaborating Claude session put it better than I had:&lt;/p&gt;&#xA;&lt;blockquote&gt;&lt;p&gt;A probe that doesn&amp;rsquo;t fail when its workload can&amp;rsquo;t reach its database isn&amp;rsquo;t a probe — it&amp;rsquo;s wallpaper.&lt;/p&gt;&lt;/blockquote&gt;&lt;p&gt;That&amp;rsquo;s the whole thing. A readiness probe answers one question: &lt;em&gt;is this pod ready to serve traffic?&lt;/em&gt; If the answer depends on a database connection and you&amp;rsquo;re not checking for that, you&amp;rsquo;re not answering the question — you&amp;rsquo;re decorating the pod spec.&lt;/p&gt;</description>
      
    </item>
    
  </channel>
</rss>
