https://docs.matildes.dev/tags/cheatsheet/Recent content in Cheatsheet on Samuel Matildes - Knowledge BaseHugoenTue, 26 May 2026 14:15:13 +0100https://docs.matildes.dev/linux/admin/linux-performance-playbook/Tue, 19 May 2026 00:00:00 +0000https://docs.matildes.dev/linux/admin/linux-performance-playbook/<p><i class="fas fa-tachometer-alt" aria-hidden="true"></i> A field-ready reference for Linux performance investigations — keep it open in a second terminal.</p> <h2 id="how-to-use-this-playbook">How to use this playbook<a class="td-heading-self-link" href="#how-to-use-this-playbook" aria-label="Heading self-link"></a></h2> <p>Linux performance problems split cleanly into <strong>two very different investigations</strong>, and picking the wrong one wastes hours:</p> <table> <thead> <tr> <th>Situation</th> <th>Track</th> <th>What you do</th> </tr> </thead> <tbody> <tr> <td>The problem is happening <strong>right now</strong> (or you can reproduce it on demand)</td> <td><strong>Track A — Live Triage</strong></td> <td>Interactive tools, sample at 1s intervals, follow the bottleneck</td> </tr> <tr> <td>The problem is <strong>intermittent / random</strong> (happens overnight, once a week, only under load you can&rsquo;t reproduce)</td> <td><strong>Track B — Background Collection</strong></td> <td>Arm continuous loggers <em>before</em> the next occurrence, then mine the logs afterwards</td> </tr> </tbody> </table> <div class="kb-callout kb-callout--rule"> <div class="kb-callout__title"> <i class="fas fa-bullseye" aria-hidden="true"></i> <span>Rule of thumb</span> </div> <div class="kb-callout__body"> If you can&rsquo;t reproduce it, <strong>do not</strong> keep staring at <code>top</code>. Stop, deploy collectors, walk away, and analyse later. Otherwise you&rsquo;ll miss the event every single time. </div> </div> <p><img src="https://docs.matildes.dev/linux/admin/linux-performance-playbook/workflow.png" alt="Decision flow: live triage vs. background collection"></p>