<?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>Onur Baydaş</title><link>https://onurbaydas.online/en/</link><description>Recent content on Onur Baydaş</description><generator>Hugo</generator><language>en-us</language><lastBuildDate>Tue, 13 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://onurbaydas.online/en/index.xml" rel="self" type="application/rss+xml"/><item><title>One User, Two UIDs: NFSv4, Kerberos, and Docker on a Domain-Joined TrueNAS</title><link>https://onurbaydas.online/posts/nfs-uid-mismatch-en/</link><pubDate>Tue, 13 Oct 2026 00:00:00 +0000</pubDate><guid>https://onurbaydas.online/posts/nfs-uid-mismatch-en/</guid><description>SSSD and TrueNAS gave the same AD user different UIDs. How that showed up as nobody, Permission denied, root:root 755, and &amp;#34;I have no name!&amp;#34;, and how we aligned them.</description></item><item><title>Docker Authorization with OPA: The Silent Fail-Open and Other Production Lessons</title><link>https://onurbaydas.online/posts/docker-authz-opa-en/</link><pubDate>Mon, 12 Oct 2026 00:00:00 +0000</pubDate><guid>https://onurbaydas.online/posts/docker-authz-opa-en/</guid><description>The plugin looks enabled, the policy is correct, yet Docker allows everything. A silent security gap I hit with opa-docker-authz in production, and its lessons.</description></item><item><title>About</title><link>https://onurbaydas.online/en/about/</link><pubDate>Mon, 01 Jan 0001 00:00:00 +0000</pubDate><guid>https://onurbaydas.online/en/about/</guid><description>&lt;p&gt;I&amp;rsquo;m Onur Baydaş, a systems and infrastructure engineer based in Turkey. Since 2024
I&amp;rsquo;ve been designing, running, and troubleshooting on-premises infrastructure: the
kind of environments where there is no managed cloud service to fall back on and
every layer, from storage to scheduling, is your responsibility.&lt;/p&gt;
&lt;h2 id="what-i-work-on"&gt;What I work on&lt;/h2&gt;
&lt;ul&gt;
&lt;li&gt;&lt;strong&gt;Virtualization and storage:&lt;/strong&gt; Proxmox VE clusters with Ceph, VMware vCenter/ESXi,
hyperconverged platforms, TrueNAS/ZFS.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;GPU infrastructure:&lt;/strong&gt; NVIDIA GPU servers for AI and research workloads, GPU
partitioning with MIG and vGPU, job scheduling with Slurm.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Multi-user platforms:&lt;/strong&gt; Shared Linux systems for many users: user isolation,
container access control, and day-to-day operations.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Container security:&lt;/strong&gt; Restricting Docker on shared machines, authorization
policies with OPA/Rego, and the gaps those policies can&amp;rsquo;t close.&lt;/li&gt;
&lt;li&gt;&lt;strong&gt;Identity and operations:&lt;/strong&gt; Samba-based Active Directory, Kerberos/NFS home
directories, centralized logging, and backup monitoring.&lt;/li&gt;
&lt;/ul&gt;
&lt;h2 id="what-i-write-about"&gt;What I write about&lt;/h2&gt;
&lt;p&gt;I write about problems I actually hit in production: the silent failures, the
misleading documentation, and what the source code says when the docs don&amp;rsquo;t.
Every post is based on a real incident, tested on a live system, and backed by
links to the source.&lt;/p&gt;</description></item></channel></rss>