<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Karl Kubelet on Medium]]></title>
        <description><![CDATA[Stories by Karl Kubelet on Medium]]></description>
        <link>https://medium.com/@kubelet?source=rss-144945e1e05d------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*3YFD-K-9T8x_DMdxIp0TQA.jpeg</url>
            <title>Stories by Karl Kubelet on Medium</title>
            <link>https://medium.com/@kubelet?source=rss-144945e1e05d------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sat, 21 Nov 2026 12:21:04 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@kubelet/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[How to run WP-CLI commands on multiple sites]]></title>
            <link>https://medium.com/@kubelet/how-to-run-wp-cli-commands-on-multiple-sites-ed6f606ed504?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/ed6f606ed504</guid>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[bash]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Mon, 31 Aug 2026 17:52:41 GMT</pubDate>
            <atom:updated>2026-08-31T17:52:41.210Z</atom:updated>
            <content:encoded><![CDATA[<p>I manage about 30 sites for my retainer clients, friends and family and WordPress projects I’m involved in. With the recent rise in security updates, it became quite tedious looking for specific plugins, forcing updates, etc. especially when the sites are scattered across different servers.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*r5DtWCSqMcNlSdzLmhMNaQ.png" /></figure><p>That’s why I wrote a wpm.sh script. It is a self-contained bash script that runs WP-CLI commands across multiple sites over SSH concurrently. It reports failures by default, but can also print the output for each individual site. Here are some examples:</p><pre># Force a minor core update<br>./wpm.sh core update --minor --force<br><br># Update all plugins<br>./wpm.sh plugin update --all<br><br># Find advanced-custom-fields<br>./wpm.sh -o &#39;plugin list | grep advanced-custom&#39; &gt; acf-sites.txt</pre><p>The sites configuration lives in the script itself, and assumes you have the same SSH key configured on each site:</p><pre>SITES=&quot;<br>karl@wpshell.com:/sites/wpshell.com/public_html<br>karl@kubeadm.org:/sites/kubeadm.org/public_html<br>#deploy@10.0.0.5:/srv/old/public<br>&quot;</pre><p>You can <a href="https://github.com/koddrio/wp-cli-multi">grab the script from GitHub</a>. It’s open source and GPLv3. Please try it out and give it a like if you find it useful!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ed6f606ed504" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[WordPress Vulnerability Scanning … Without a Plugin]]></title>
            <link>https://medium.com/@kubelet/wordpress-vulnerability-scanning-without-a-plugin-41d1b11b9efe?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/41d1b11b9efe</guid>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[sysadmin]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[security]]></category>
            <category><![CDATA[web-development]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Tue, 27 Jan 2026 13:33:47 GMT</pubDate>
            <atom:updated>2026-06-03T18:10:21.456Z</atom:updated>
            <content:encoded><![CDATA[<h3>WordPress Vulnerability Scanning … Without a Plugin</h3><p>This is probably my very first “without a plugin” post. I don’t dislike plugins in general, and I do believe that dumping something in functions.php for a “without a plugin” post is stupid. But this is different.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*BG40be98JXF9MaxV6IO25w.png" /><figcaption>Wordfence CLI vulnerability scan</figcaption></figure><h3>Security Plugins</h3><p>I’ve seen so many security plugins over the years, from ones hiding the WordPress version and the readme.html file, to ones matching plugin versions against a vulnerability database, or pattern matching for known malware signatures, which might be quite useful.</p><p>Then again, the problem I have with these plugins is the fact that they’re doing (almost) everything in PHP, inside of an HTTP request. This is extremely inefficient use of resources. While PHP can technically do threading with pthreads it’s really not that useful in the context of an HTTP request and PHP-FPM or mod_php.</p><p>In addition to that, these security plugins load stuff on every single request, especially the commercial ones and their “lite” versions. The scan itself may happen once every night during a maintenance window, why should the remaining 99.99% of requests suffer from the bloat?</p><p>Also, what if the WordPress site is already infected? What if it’s been compromised? What if it’s offline, or a backup, or behind a “block all” firewall?</p><p>I want my malware and vulnerability scanning to continue working, regardless of the state of my site. I want to run scans on backups, which helps identify when a file was first infected. I want to run scans on dozens of sites on a single server, without having to hop and juggle 12 wp-admin tabs. Is that too much to ask?</p><p>Well, no.</p><h3>Wordfence CLI</h3><p>As much as I dislike the Wordfence plugin and their business model, <a href="https://github.com/wordfence/wordfence-cli">the CLI utility</a> is an absolute gem. It’s free and open source, released under the GPLv3, supports multi-threaded scanning for malware and vulnerabilities.</p><p>There is a catch of course. It uses the same vulnerability and malware database that the plugin version uses, and has a 30-day delay on malware signatures unless you pay or the premium version. They may one day decide to add a delay to vulnerabilities, or paywall the entire thing, but until that day arrives, this is by far the best, and most affordable approach to vulnerability and malware scanning for WordPress.</p><p>Installing is not as easy as a plugin of course. On the <a href="https://github.com/wordfence/wordfence-cli/releases">releases page</a> there are pre-compiled binaries for Linux, and packages for Debian/Ubuntu and Red Hat-based distributions. I’m running Ubuntu on my test server, so I’ll use the .deb along with some dependencies:</p><pre>sudo apt install -y python3-pymysql libpcre3<br>wget https://github.com/wordfence/wordfence-cli/releases/download/v5.0.2/wordfence.deb<br>sudo dpkg -i wordfence.deb<br>rm wordfence.deb</pre><p>Then, after a quick setup and free license agreement, I can easily run a vulnerability and a malware scan across all my sites:</p><pre>wordfence vuln-scan /sites<br>wordfence malware-scan /sites</pre><p>This will recursively search my /sites directory for anything that looks like a WordPress site and perform the scan. The vulnerability scan is pretty quick as it only compares versions. The malware scan has to read and pattern-match every relevant file, so it is much slower and more CPU intensive, but unlike the PHP stuff, it can utilize multiple CPUs and CPU cores very efficiently!</p><p>Using the --email flag and a system cron configuration, I can run this on a schedule and have the CLI send me (or my client) <a href="https://wpshell.com/lesson/postfix-relay/">an email report</a> if something is found.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*P2CYJ5_7Xy0ZBuRcuKeNhg.png" /><figcaption>Wordfence CLI vulnerability email report</figcaption></figure><p>I manage quite a few client sites and discovering <a href="https://www.wordfence.com/products/wordfence-cli/">Wordfence CLI</a> was a game changer for me. I’ve been very happy with the free version for many years, though a commercial subscription is also available if you want real-time signatures.</p><p>I have an automated nightly scan across all sites, but I also do ad-hoc scans when migrating sites from another hosting provider.</p><p>I’m currently working on integrating this into a simple monitoring system which I already use for alerts. If you’re interested in self-hosting WordPress, don’t forget to <a href="https://wpshell.com/">check out my course</a>.</p><p>Stay safe!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=41d1b11b9efe" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Dedicated vs VPS for WordPress]]></title>
            <link>https://medium.com/@kubelet/dedicated-vs-vps-for-wordpress-6331df0fd44e?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/6331df0fd44e</guid>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[php]]></category>
            <category><![CDATA[web-hosting]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Wed, 19 Nov 2025 09:58:38 GMT</pubDate>
            <atom:updated>2026-06-03T18:10:50.374Z</atom:updated>
            <content:encoded><![CDATA[<p>I ran some simple benchmarks last week to compare dedicated servers vs VPS for WordPress performance. The goal was to see what you could get for about $200/mo in hosting costs from AWS, Google Cloud and DigitalOcean, versus a few dedicated server specs from Cherry Servers.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><p>I used my favorite load testing tool Oha. It’s simple, lightweight, runs on a toaster and doesn’t require hours of configuration like other more complex tools, such as Locust or Grafana k6.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1005/0*LscjsVEjhcavKg89.png" /></figure><p>I’ve also added a $1.3k/mo VM for comparison, which surprisingly performed the same as the EPYC 7543, and outperformed the Ryzen 9900X. However, the per-dollar performance value was ridiculous.</p><p>This is a raw performance comparison, not overall value, flexibility, scalability, storage and network redundancy and all the extra bells and whistles you typically don’t get with a dedicated server.</p><p>As part of this testing I’ve also compared some WordPress object and page caching plugins. <a href="https://wpshell.com/lesson/benchmarks/">Full story here</a>, with raw data and results.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6331df0fd44e" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[MySQL ZFS Snapshot w/ Binlog]]></title>
            <link>https://medium.com/@kubelet/mysql-zfs-snapshot-w-binlog-0cc6362418ea?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/0cc6362418ea</guid>
            <category><![CDATA[software-development]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[mysql]]></category>
            <category><![CDATA[linux]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Wed, 15 Oct 2025 15:56:08 GMT</pubDate>
            <atom:updated>2026-06-03T18:11:11.674Z</atom:updated>
            <content:encoded><![CDATA[<p>If you’re doing ZFS snapshots of your entire /var/lib/mysql directory, then great, you’re doing it right! That will create a consistent snapshot, which you can easily restore to. Even if something hasn’t been flushed to disk yet, there is a record of it in the MySQL/MariaDB redo logs.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*18wYPWex9EpjTE7Y" /><figcaption>Photo by <a href="https://unsplash.com/@abject?utm_source=medium&amp;utm_medium=referral">benjamin lehman</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>You don’t need to stop your server, you don’t need to lock tables or anything like that. Any inconsistencies will be sorted out during MySQL and InnoDB recovery phase when it starts.</p><p>Brilliant.</p><p><strong>Except</strong> if you have a binary log. This changes everything. For your snapshot to be aligned with a specific position in your binary log, you have to ensure that everything is flushed to disk when taking the snapshot. You must FLUSH TABLES WITH READ LOCK and hold the lock while making the snapshot. The binlog file/position shown while the lock is held will be accurate for that snapshot.</p><p>Don’t get me wrong. You can still recover from a live snapshot, however your binary log will be of very limited use. For <strong>point in time recovery</strong> in particular, you’ll want to start from a snapshot, and replay any necessary statements from the binary log <strong>starting at the position</strong> of the snapshot, to the destinatin point in time.</p><p>If that starting position is wrong, you’re screwed. That is why you want your binlog aligned. That’s why you need to flush and hold the read lock while snapshotting.</p><h3>Manual Snapshot</h3><p>You can do this manually of course. Have two terminals or sessions open, one with a MySQL shell, acquiring and holding the lock:</p><pre>FLUSH TABLES WITH READ LOCK;<br>SHOW MASTER STATUS;</pre><p>Make note of the binary log filename and position. Leave the shell open while using a second session to ZFS snapshot:</p><pre>zfs snapshot tank/mysql@2025-10-15</pre><p>Then go back to your first terminal and run:</p><pre>UNLOCK TABLES;</pre><p>Now, together with the filename and position from SHOW MASTER STATUS , your ZFS snapshot will be perfectly aligned, allowing you use the binary log for point-in-time restoration between this snapshot and the next.</p><p>Note that while the lock is held, all writes to the MySQL database are blocked. They will resume the moment you release the locks, but some of them might time out. This is why it’s a great idea to script all of this, to minimize the time the lock is held.</p><h3>Scripted Snapshot</h3><p>This is where it gets a bit interesting, especially in Bash, where concurrency is, well, let’s just say not a first class citizen. However, it is still all perfectly doable with a FIFO , a background process, and a long-held file descriptor:</p><pre>#!/bin/bash<br>set -euo pipefail<br><br>TIMESTAMP=$(date +%s)<br>FIFO=&quot;/tmp/mariadb.fifo&quot;<br>rm -f $FIFO<br>mkfifo &quot;$FIFO&quot;<br><br># Acquire and hold the lock<br>mariadb -uroot -B -n -N -q -e &quot;<br>    SELECT CONNECTION_ID();<br>    FLUSH TABLES WITH READ LOCK;<br>    SHOW MASTER STATUS;<br>    SELECT SLEEP(30);<br>    UNLOCK TABLES;&quot; \<br>    &gt;&quot;$FIFO&quot; &amp;<br><br># Keep fifo open<br>exec {RFD}&lt;&quot;$FIFO&quot;<br><br># Read from fifo<br>IFS= read -r -t 1 THREAD_ID      &lt;&amp;$RFD<br>IFS= read -r -t 10 MASTER_STATUS &lt;&amp;$RFD<br><br># Drain fifo in bg<br>cat &lt;&amp;$RFD &gt; /dev/null &amp;<br><br># Create snapshot<br>zfs snapshot &quot;tank/mysql@$TIMESTAMP&quot;<br><br>mariadb -uroot -B -n -N -q -e &quot;KILL $THREAD_ID;&quot;<br><br># Close fifo fd<br>exec {RFD}&lt;&amp;-<br><br>echo &quot;Snapshot created: $TIMESTAMP $MASTER_STATUS&quot;</pre><p>This script acquires and holds a lock while writing back the connection ID and the output from SHOW MASTER STATUS to the named pipe. It keeps the pipe open to ensure the MySQL session stays open while taking the ZFS snapshot. It then starts a new MySQL process that kills the running connection, and finally closes the named pipe.</p><p>There’s a lot that can be improved in this script, particularly around error handling. For example, it would be a great idea to record the PIDs of both background processes and kill them in a trap. You might also want to check the initial mariadb process is still running after taking the snapshot, i.e. it hasn’t timed out, hasn’t been interrupted, OOM killed, etc. I’ll leave all that up to you.</p><p>Good luck, and don’t forget to ship your backups off-site!</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=0cc6362418ea" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why WordPress SMTP plugins suck (and what you should use instead)]]></title>
            <link>https://medium.com/@kubelet/why-wordpress-smtp-plugins-suck-and-what-you-should-use-instead-f0662dc40052?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/f0662dc40052</guid>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[email]]></category>
            <category><![CDATA[web-hosting]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Tue, 30 Sep 2025 13:57:38 GMT</pubDate>
            <atom:updated>2026-06-03T20:06:07.694Z</atom:updated>
            <content:encoded><![CDATA[<p>Every WordPress admin has run into it: the dreaded <strong>“Emails aren’t sending”</strong> problem. You install WooCommerce, enable customer notifications, maybe even set up password resets , and then nothing shows up in the inbox.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*ITRLxmni4ILvpxhvQ36lSg.png" /></figure><p>The first thing you’ll find when you Google the issue? A swamp of WordPress SMTP plugins. Post-SMTP, WP Mail SMTP, Easy WP SMTP, take your pick. They all promise the same thing: <em>fix your WordPress email problems by relaying everything through Gmail, SendGrid, or whatever SMTP service you can get your hands on.</em></p><p>Sounds like the perfect fix, right? Nope.</p><h3>The dirty truth about SMTP plugins</h3><p>On paper, these plugins look simple: they hijack WordPress’ wp_mail() call, open an SMTP connection to some provider, and push the message out. In practice, they suck. Here’s why:</p><ol><li><strong>They block your request.</strong><br>Sending an email over SMTP means opening a network connection, negotiating TLS, authenticating, transmitting the whole message, waiting for a response… all inside the same PHP request that’s handling your customer’s checkout or password reset. That’s hundreds of milliseconds at best, many seconds at worst. Your users are left staring at a spinning wheel while WordPress negotiates with Gmail’s mail servers.</li><li><strong>They fail silently.</strong><br>If the upstream SMTP service is down, slow, or drops the connection, your plugin will usually just throw a PHP error or timeout. No retries, no queueing, no backoff. The email is gone forever.</li><li><strong>They eat your server resources.</strong><br>A single slow SMTP transaction ties up a PHP worker. Multiply that by dozens of concurrent WooCommerce checkouts, and you’re looking at workers wasted on email instead of serving pages.</li><li><strong>They’re unnecessary bloat.</strong><br>All they really do is replicate what a proper mail system on the server already knows how to do, except worse.</li></ol><h3>What you should be using: Postfix + a relay</h3><p>Linux already solved this problem decades ago. That’s what <strong>Postfix</strong> (or Exim, or OpenSMTPD) is for.</p><p>Here’s how it should work:</p><ul><li>WordPress calls wp_mail().</li><li>PHP hands the message to the system’s sendmail binary.</li><li>That binary just writes the message to a local queue (postdrop).</li><li>Done. Your PHP request finishes instantly.</li></ul><p>From there, Postfix takes over <em>asynchronously</em>. It will retry if the destination server is busy. It will queue messages until the network comes back. It will batch, back off, and generally do all the things SMTP plugins will never do.</p><p>You can configure Postfix to send mail directly to the recipient’s mail server, or, better yet, relay through a service like Amazon SES, Postmark, or Mailgun. That way you get professional-grade deliverability while still keeping the reliability and efficiency of Postfix’s local queue.</p><h3>The freedom of running your own server</h3><p>Managed WordPress hosts often lock down outbound email, which is why SMTP plugins became popular in the first place. If you’re on a VPS or a dedicated server, you don’t need them. You’ve got full control. Install <a href="https://www.postfix.org/">Postfix</a>, configure it to relay, and throw every SMTP plugin in the trash. Here’s an <a href="https://www.linuxbabe.com/mail-server/postfix-smtp-relay-ubuntu-sendinblue">old but very good guide</a> on Postfix relay configuration. For WordPress specifically, <a href="https://wpshell.com/lesson/understanding-email/">start here</a>.</p><p>Your WordPress sites (and any other services on the box , monitoring tools, cron jobs, whatever) will all share the same robust email pipeline without a single extra plugin.</p><p>WordPress SMTP plugins are a band-aid at best and a performance-sapping, failure-prone crutch at worst. If you actually care about reliable email delivery, skip the plugins and set up Postfix.</p><p>Your users don’t want to wait for SMTP handshakes, and you don’t want to explain why their password reset never arrived.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=f0662dc40052" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[You’ve been doing phpMyAdmin wrong]]></title>
            <link>https://medium.com/@kubelet/youve-been-doing-phpmyadmin-wrong-c2680653cd72?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/c2680653cd72</guid>
            <category><![CDATA[web-hosting]]></category>
            <category><![CDATA[php]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[wordpress]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Fri, 19 Sep 2025 20:06:57 GMT</pubDate>
            <atom:updated>2026-06-03T20:08:18.518Z</atom:updated>
            <content:encoded><![CDATA[<p>If you’re still uploading phpMyAdmin to your web root and calling it a day, you&#39;re setting yourself up for pain, or worse, a breach. phpMyAdmin is one of the most scanned-for (by bots) applications on the Internet, and has over 270 public CVE (vulnerability) records.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><p>It needs to be private, and it needs to be up to date, automatically.</p><p>Here’s how to do it right on Ubuntu:</p><ul><li>No sketchy zip downloads</li><li>No insecure web exposure</li><li>No default settings left behind</li></ul><p>Instead, we’ll use:</p><ul><li>Ubuntu’s native package</li><li>A dedicated php-fpm pool</li><li>A locked-down nginx server block that only listens locally</li><li>And an SSH tunnel so you, and only you, can access it</li></ul><p>Let’s fix your setup.</p><h3>Step 1: Install phpMyAdmin using apt</h3><p>Ubuntu and most other Debian-based distributions, have an official phpmyadmin package. Yes, this comes with a dependency of about 50 other individual packages, including PHP modules and JavaScript libraries, and that is okay.</p><p>At least it doesn’t have a node_modules directory, and you don’t have to manually build it with Composer.</p><pre>sudo apt update<br>sudo apt install phpmyadmin</pre><p>This thing tires to be way too helpful. When prompted:</p><ul><li>Skip automatic web server configuration (do not select any server, or select “None” if it’s an option)</li><li>Select No to configure the database</li></ul><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*xjFDrocizuTxOY-wQsKG_A.png" /></figure><p>This installs the PHP application and config files, but doesn’t yet expose any of it on the web. Application files are in /usr/share/phpmyadmin (with a bunch of other deps in /usr/share), pre-generated config files are in /etc/phpmyadmin. There’s nothing to configure really, but if you really need to, you can drop PHP files in /etc/phpmyadmin/conf.d and they’ll be picked up automatically.</p><h3>Step 2: Configure a dedicated php-fpm pool</h3><p>You don’t want phpMyAdmin to run as part of your main PHP-FPM pool, or a pool that belongs to a specific site. You want a custom pool, with some custom rules, and dedicated workers, which you’ll be able to access even if your other sites are being DoS’d.</p><pre>[phpmyadmin]<br>user = www-data<br>group = www-data<br>chdir = /usr/share/phpmyadmin<br><br>listen = /var/run/phpmyadmin.sock<br>listen.owner = www-data<br>listen.group = www-data<br><br>pm = ondemand<br>pm.max_children = 4<br><br>php_admin_value[open_basedir] = /usr/share:/etc/phpmyadmin:/tmp<br>php_admin_value[memory_limit] = 512M</pre><p>With open_basedir we restrict to just the directories we expect. You can also have a custom memory_limit or other PHP-related settings here for phpMyAdmin, if you’re planning to do big imports/exports, etc.</p><p>This file should go to /etc/php/8.4/fpm/pool.d/phpmyadmin.conf (replace 8.4 with your PHP version). Restart PHP-FPM to read the new pool definition:</p><pre>sudo systemctl restart php8.4-fpm.service</pre><h3>Step 3: Create a local-only nginx server block</h3><p>Do not run phpMyAdmin as part of an existing, publicly accessible server block. We’ll want a dedicated server block, doesn’t need a server name, doesn’t even need HTTPS/TLS, because we’re staying indoors.</p><pre>server {<br>    listen 127.0.0.1:8080 default_server;<br>    server_name _;<br><br>    root /usr/share/phpmyadmin;<br>    index index.php;<br><br>    location ~ \.php$ {<br>        include snippets/fastcgi-php.conf;<br>        fastcgi_pass unix:/var/run/phpmyadmin.sock;<br>    }<br>}</pre><p>We bind to port 8080 on 127.0.0.1 so even if our systems admin (i.e. we) made a mistake and exposed that port publicly, Nginx willl not accept any connections on it.</p><p>Drop this file into /etc/nginx/conf.d or /etc/nginx/sites-enabled and reload the service:</p><pre>sudo systemctl reload nginx.service</pre><p>Now comes the fun part.</p><h3>Step 4: Access phpMyAdmin over SSH tunnel</h3><p>phpMyAdmin is up and running, but it’s listening on a local port on the server, with no way to access it externally. However, using one of my favorite techniques, <strong>SSH Tunneling</strong>, we can create a tunnel between our server, and our home/work computer.</p><pre>ssh -N -L 8080:127.0.0.1:8080 karl@my.server.tld</pre><p>This works on Windows, Mac and Linux. It creates a tunnel on port 8080, to a remote location of 127.0.0.1:8080 on the target my.server.tld server. You’ll need to authenticate of course, but once the tunnel is up and running, you can simply point your browser to http://127.0.0.1:8080 to access your remote phpMyAdmin instance.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*L1D_Ab4oNYVf7e7cnxVVQA.png" /></figure><h3>Why this is better</h3><ul><li><strong>No public exposure</strong>: Attackers can’t scan what they can’t see.</li><li><strong>Isolated PHP pool</strong>: Limits blast radius if something breaks or gets exploited. Even if main pools are overwhelmed, phpMyAdmin still works like a charm.</li><li><strong>Package-managed</strong>: Easier, potentially automatic updates.</li></ul><p>You will need to have <a href="https://help.ubuntu.com/community/AutomaticSecurityUpdates">Automatic Security Updates</a> enabled to ensure your phpMyAdmin is updated when a new security release is available. However, even if you don’t, guess what? Nobody except you can get to it anyway.</p><h3>Bonus: Lock it down even more</h3><p>Want extra paranoia points?</p><ul><li>Add basic auth to your nginx config.</li><li>Use an SSH key, limit SSH access by IP, add 2-factor.</li><li>Use fail2ban to catch brute force attempts on SSHd.</li><li>Use fail2ban to disable phpMyAdmin (remove Nginx config and reload) after 3 incorrect password attempts on phpMyAdmin.</li></ul><p>Next time you see an exposed phpmyadmin directory lying around in a public_html folder, delete it immediately and go to step 1. If you’re looking to learn more about hosting stuff, specifically around WordPress, check out my <a href="https://wpshell.com/">self-hosting WordPress course</a>.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c2680653cd72" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Secure your Nginx origin with Cloudflare (AOP + Origin Certificates)]]></title>
            <link>https://medium.com/@kubelet/secure-your-nginx-origin-with-cloudflare-aop-origin-certificates-eee28cc71594?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/eee28cc71594</guid>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[nginx]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[web-hosting]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Fri, 12 Sep 2025 12:14:02 GMT</pubDate>
            <atom:updated>2026-06-03T20:10:27.184Z</atom:updated>
            <content:encoded><![CDATA[<p>In this lesson you’ll learn how to fully integrate Cloudflare with your Nginx server. We’ll cover how to block direct access to your origin, enforce mutual TLS with Cloudflare using Authenticated Origin Pulls, replace self-signed certificates with Cloudflare-issued origin certificates, and configure Nginx to log the real visitor IPs.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><p>Even though we currently have a domain set up on Cloudflare, our current configuration allows anyone on the Internet to access our HTTPS service directly, either by using the IP address of our server, or by guessing our origin host name.</p><p>This presents several challenges, primarily related to security and performance. For example, direct visitors will bypass any WAF (web application firewall) and caching rules configured on Cloudflare, including rate limiting and captcha challenges.</p><h3>Allowing only Cloudflare traffic</h3><p>There are multiple ways to lock our server down, to accept HTTPS traffic from Cloudflare only. For example, we can use the <a href="https://wpshell.com/lesson/firewall/">UFW firewall</a> to block all HTTPS traffic except traffic coming <a href="https://www.cloudflare.com/ips-v4">from Cloudflare’s IPs</a>. While this is a viable option, it does require maintaining lists of IPs.</p><p>Another way is using Cloudflare’s <a href="https://developers.cloudflare.com/ssl/origin-configuration/authenticated-origin-pull/">AOP</a> (Authenticated Origin Pulls). This asks Cloudflare to present a client certificate while communicating with the origin server. If we can verify this certificate, we can be sure that the request is coming from Cloudflare.</p><p>Let’s enable this option for our domain in the Cloudflare Dashboard under the SSL/TLS — Origin Server section:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*T1lNrJgH6OCbtkFE.png" /></figure><p>However, simply enabling it is not enough. We have to validate this certificate in Nginx.</p><h3>Nginx configuration</h3><p>First let’s download the Cloudflare AOP certificate and add it to our configs repository:</p><pre>cd /config/ssl<br>sudo curl -o cloudflare-aop.pem https://developers.cloudflare.com/ssl/static/authenticated_origin_pull_ca.pem</pre><p>Next, let’s update our /config/nginx/uncached.org.conf file to verify the client&#39;s request against this certificate inside our server block:</p><pre>ssl_client_certificate /config/ssl/cloudflare-aop.pem;<br>ssl_verify_client on;</pre><p>Save and reload your Nginx configuration:</p><pre>sudo systemctl reload nginx.service</pre><p>At this point, we should no longer be able to access our HTTPS service without presenting a client certificate. We can verify this using cURL locally:</p><pre>curl -v --insecure -H &quot;Host: uncached.org&quot; https://127.0.0.1/test.php</pre><p>This should now respond with a 400 error code, and a &quot;No required SSL certificate was sent&quot; message.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*bM8YnYpxKJIHSCy8.png" /></figure><p>This is because we have not sent a client certificate at all. We can also make sure that the Cloudflare certificate is actually validated, by sending a request with a different client certificate instead.</p><p>If you still have your self-signed certificate and key lying around, you can test with those:</p><pre>sudo curl \ <br>  --cert /config/ssl/uncached.org.cert.pem \ <br>  --key /config/ssl/uncached.org.key.pem \ <br>  -v --insecure -H &quot;Host: uncached.org&quot; \ <br>  https://127.0.0.1/test.php</pre><p>This will send the wrong client certificate to Nginx, and it should respond with code 400 and a &quot;SSL certificate error&quot;. Viewing the domain through Cloudflare (in a browser for example) should still work just fine:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*I6Dss5Ci1ErUy1Wx.png" /></figure><p>Now that we have a way to fully trust requests coming from Cloudflare, let’s also make sure Cloudflare can fully trust responses coming from our server.</p><h3>Origin certificates</h3><p>Previously when we set up our first site in Nginx, we generated a self-signed certificate to serve over HTTPS. Let’s now replace that certificate with one generated by and trusted by Cloudflare:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*oO4eaSa5nm93zetK.png" /></figure><p>You can generate a new certificate in the <em>SSL/TLS — Origin Server — Origin Certificates — Create Certificate</em> section in your Cloudflare dashboard. You’ll be able to select the host names to add to the certificate too. I’ve added uncached.org and a wildcard *.uncached.org to mine, which covers www.uncached.org and other subdomains I may wish to add in the future.</p><p>Copy the resulting <em>Origin Certificate</em> and <em>Private Key</em> to your server configs, replacing the old self-signed ones. Mine are in/config/ssl/uncached.org.cert.pem and /config/ssl/uncached.org.key.pem respectively. Note that the permissions on the private key file should be 0600, owned by root:</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*_Yi97y7vPxoHhfhg.png" /></figure><p>You can double-check your certificate contents (issuer, dates, etc.) with openssl:</p><pre>openssl x509 -in uncached.org.cert.pem -noout -text</pre><p>Then reload your Nginx service using:</p><pre>sudo systemctl reload nginx.service</pre><p>Now you should be able to update your Cloudflare SSL/TLS settings and set the encryption mode to <em>Full (Strict)</em>. This option enforces validation of the newly generated origin certificates.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*RbPyPvDH3ivYK5pM.png" /></figure><p>If you’ve done everything correctly, you should still be able to access your site via Cloudflare from a browser window.</p><h3>HTTP to HTTPS redirect</h3><p>Our Nginx server doesn’t even listen on the insecure HTTP port, however with the default settings, Cloudflare will still attempt to send traffic to port 80 if the user requested an HTTP resource.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*cVacE9MzMYfot-RV.png" /></figure><p>We can address this by redirecting all HTTP traffic to HTTPS at the edge. You can find this setting in <em>SSL/TLS — Edge Certificates — Always Use HTTPS</em> in the Cloudflare dashboard. You can optionally turn on HSTS there too and configure your TLS versions if needed.</p><h3>Nginx Real-IP</h3><p>Being behind Cloudflare, there’s one last thing we need to resolve: visitors’ real IPs. If you look at your /var/log/nginx/access.log you will see traffic coming from certain IP addresses. You can use whois to get more information on those IPs and you&#39;ll soon find that all of them belong to Cloudflare.</p><p>This is not surprising, because all this traffic is <em>indeed</em> coming from Cloudflare. However, this will be a problem for any application security we implement going forward, including any audit and logging. We’ll want the real end user’s IP addresses for those purposes.</p><p>This is where Nginx’s <a href="https://nginx.org/en/docs/http/ngx_http_realip_module.html">realip_module</a> comes in. It allows us to set the client’s IP address from a header, which is propagated down to logs, FastCGI (and PHP) requests and more.</p><p>Cloudflare sends us the visitor’s IP address in a CF-Connecting-IP header. Let&#39;s update our site configuration (mine is in/config/nginx/uncached.org.conf) to include this in the server block:</p><pre>set_real_ip_from 0.0.0.0/0;<br>real_ip_header CF-Connecting-IP;</pre><p>The first directive tells Nginx that we can trust the real IP header from any source. We can only do this because we know that the <em>only source</em> for our traffic in this server block is Cloudflare, due to the AOP configuration we implemented earlier. Do not do this if you are not enforcing AOP, otherwise you allow anyone to spoof IPs.</p><p>The next directive tells Nginx which header to derive the IP address from. After saving your configuration and reloading your Nginx service, you should now see your ISP’s or VPN IP address in the access log.</p><h3>Reusable configuration</h3><p>If you’re planning to run multiple WordPress sites from the same dedicated or virtual server, you’ll find yourself copying chunks of configuration for every site you configure. I encourage you to find reusable things in the configuration, especially in Nginx, and provide these as snippets you could include instead.</p><p>For example, the client certificate, client certificate verification and the realip configuration for Cloudflare are not going to change, so you could move them to a /config/nginx/cloudflare.conf and include that file inside the server block to make it a bit cleaner.</p><p>This will also prevent you from making a mistake in a future configuration, by leaving out ssl_verify_client for example, allowing non-Cloudflare traffic through.</p><p><em>Originally published at </em><a href="https://wpshell.com/lesson/cloudflare-config/"><em>https://wpshell.com</em></a><em>.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=eee28cc71594" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[CPU core count and frequency]]></title>
            <link>https://medium.com/@kubelet/cpu-core-count-and-frequency-a621c3cfcc86?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/a621c3cfcc86</guid>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[php]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[linux]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Thu, 11 Sep 2025 15:37:12 GMT</pubDate>
            <atom:updated>2026-06-03T20:11:38.375Z</atom:updated>
            <cc:license>http://creativecommons.org/licenses/by/4.0/</cc:license>
            <content:encoded><![CDATA[<p>Both virtual machines and bare metal servers come in different shapes and sizes. When comparing one offering to another, we’re typically looking at things like memory, disk size and CPU. Without looking into more detail, however, it’s easy to make false assumptions, especially when such details are abstracted and hidden away.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><p><em>This post </em><a href="https://wpshell.com/lesson/cpu-cores-frequency/"><em>first appeared</em></a><em> on WP Shell, a full step-by-step guide on </em><a href="https://wpshell.com/"><em>self-hosting WordPress</em></a><em> on your own dedicated or virtual server.</em></p><p>In this post we’re going to look at CPUs and some tricky terminology around them like cores, threads, virtual CPUs, frequencies and more, and how it all relates to running a WordPress application.</p><h3>CPUs, cores, threads and frequency</h3><p>A CPU, or processor, is a physical microchip plugged into the motherboard of a computer. While there are plenty of dual- and even quad-socket motherboards, most mass-market hosting offerings are centered around single-socket computers.</p><p>A CPU <em>core</em> is part of a CPU which does the actual processing. Modern-day CPUs typically have between 8 and 128 cores, allowing us to process data in parallel.</p><p>A CPU <em>thread</em> is a further extension of that, which allows one physical CPU core to act like two separate <em>logical</em> cores. For example, a 16-core processor with SMT (Hyper-Threading, etc.) will show 32 available cores in the operating system. This generally improves parallel processing even more.</p><p>The CPU <em>frequency</em>, or its <em>clock speed</em>, is the rate at which the CPU executes instructions. It’s not a very useful metric when comparing CPUs across different families, generations and architectures, as it is not the only factor in determining performance.</p><h3>Dedicated servers</h3><p>When it comes to dedicated server specs, you’ll usually know the exact model of the CPU in your server, which gives you the exact core count, frequency, whether multi-threading is available and other details. Even if the hosting provider doesn’t publish these numbers, you can always look them up on the processor vendor’s website.</p><p>While on a server, you can identify the CPU model, architecture, cache sizes, core count and various other things using the commands:</p><pre>lscpu<br>lscpu -e<br>cat /proc/cpuinfo</pre><p>Comparing processors is quite easy. I always use the <a href="https://www.cpubenchmark.net/multithread/server">cpubenchmark.net</a> website for single-threaded and multi-threaded CPU comparison. The Geekbench <a href="https://browser.geekbench.com/processor-benchmarks">processor benchmarks</a> are also a useful resource.</p><h3>Virtual machines</h3><p>With VMs it’s a lot muddier, and this is where the marketing departments absolutely love to do their dirty work. You can use the commands listed above and you’ll be lucky to get some information. A lot of the time it’ll be quite obfuscated.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*amBPiAT4WTHrtNVz.png" /></figure><p>Resource allocation with virtual machines is as flexible as it gets, and completely controlled by the provider. What this means is that whatever providers publish on their pricing page is basically meaningless.</p><p>Some will reference Cores, others will say Threads, some might say vCPUs, ECUs, CPU units, sometimes even outright CPUs. This is quite misleading, and they might as well call them carrots. However, these carrots give us two somewhat useful pieces of information:</p><ul><li>The number of carrots advertised is <em>usually</em> the number of logical CPUs we’ll see in the guest OS</li><li>4 carrots is typically twice as fast as 2 carrots within the same vendor</li></ul><p>Do note that some vendors will also have compute-optimized carrots, high-frequency carrots, and others. What this means is that we get more CPU allowance on the same number of carrots. This brings us to one of the most important performance factors when it comes to virtual machines.</p><h3>CPU quota or allowance</h3><p>If a CPU core is capable of performing X amount of work every second, an allowance or CPU quota of 50% will allow our virtual machine to perform only <em>half</em> of X every second. An allowance of 5% gives us only 1/20th of the full core capacity.</p><p>This is a very simplified explanation. In reality things often get a lot more complex, with variable allowances, throttling, baseline reservations, burst capacity and even CPU credit systems where your VM accumulates “credits” when it’s idle, and can then “spend” these credits to use some CPU. The <a href="https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/burstable-credits-baseline-concepts.html">entire T-lineup on AWS</a> (t2, t3, t4g instances) is built on such a system.</p><p>Some cloud and VM providers have a “dedicated CPU” offering, where they guarantee 100% allowance/quota for each listed core, usually at a significantly more expensive price. However, most common offerings are shared-CPU, meaning you get a small slice, and a slightly larger slice with “compute optimized” offerings.</p><p>The actual size of that slice is almost never publicized, as vendors want to retain the ability to change these without notice for rebalancing, cost optimization and more.</p><p>This makes it increasingly hard to compare an offering from one vendor to a similar offering from another vendor. Furthermore, provisioning a cloud instance is a bit of a lottery in itself too, as you may get extremely lucky and placed on a freshly deployed server with very low density and silent neighbors. Or you can get extremely unlucky, and end up on an old, soon-to-be decommissioned, oversubscribed, max-density server with a bunch of spammers.</p><p>Luckily, you can always run a Geekbench or an HTTP benchmark on two servers and find out which gives you better performance. You can find existing benchmarks online too for some overall guidance. Do note that synthetic benchmarks (Geekbench, PressMark) are directional, but nothing beats load testing an actual WordPress site to determine true performance.</p><h3>CPUs for WordPress</h3><p>There is no one-size-fits-all unfortunately. WordPress is a single-threaded application (a single request can only run on one core), and depending on the plugins, themes and overall configuration, can easily be CPU-bound or IO-bound, sometimes depending on the time of day.</p><p>It’s a PHP application which typically runs behind a multi-threaded or multi-process manager, such as PHP-FPM. Combine that with Nginx and MySQL/MariaDB, and you can definitely make good use of multiple cores, but you don’t want to sacrifice single-core performance too much.</p><p>A good rule of thumb is looking at the cache hit-rate. With 90% and more it’s safe to sway toward more CPU cores for a higher worker count and concurrency. With low cache hit rates and highly dynamic sites, it’s often more beneficial to invest in better single-core performance.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a621c3cfcc86" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[On Blocking xmlrpc.php]]></title>
            <link>https://medium.com/@kubelet/on-blocking-xmlrpc-php-aceecb09ab7b?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/aceecb09ab7b</guid>
            <category><![CDATA[security]]></category>
            <category><![CDATA[wordpress]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Thu, 04 Sep 2025 09:26:33 GMT</pubDate>
            <atom:updated>2026-06-03T20:12:43.637Z</atom:updated>
            <content:encoded><![CDATA[<p>Just landed another customer for a performance and security audit, and they’re missing xmlrpc.php from their installation :(</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/">Koddrio Cloud</a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*9_lV_toz0MYIlvH3" /><figcaption>Photo by <a href="https://unsplash.com/@fabioha?utm_source=medium&amp;utm_medium=referral">fabio</a> on <a href="https://unsplash.com/?utm_source=medium&amp;utm_medium=referral">Unsplash</a></figcaption></figure><p>If you run WordPress, chances are you’ve heard advice to block xmlrpc.php for “security.” It sounds simple: shut the door and the bad guys stay out. But in reality, blocking xmlrpc.php often breaks more than it protects.</p><p><strong>In my experience:</strong></p><ol><li><strong>It’s not insecure. </strong>The XML-RPC API isn’t a flaw, it’s a core feature of WordPress. Attacks that target it (like brute force login attempts) are just that: attacks. The real issue is weak security practices, not the file itself.</li><li><strong>It powers some essential integrations. </strong>Popular services rely on xmlrpc.php: Jetpack, the WordPress mobile app, remote publishing tools, even some plugins. Block it, and you quietly disable functionality you or your users might actually want.</li><li><strong>There are better defenses.</strong> If you’re worried about brute force or DoS attacks, don’t disable XML-RPC. Instead: Use strong passwords and two-factor authentication, Limit login attempts using tools like <a href="https://wpshell.com/lesson/fail2ban-ssh/">fail2ban</a>, deploy a <a href="https://wpshell.com/lesson/firewall/">firewall</a> or rate-limiting at the server level.</li><li>Blocking xmlrpc.php is like putting tape over your car’s gas cap because you’re worried about fuel theft. You might stop the thief, but you also stop yourself from driving.</li></ol><p><strong>Bottom line:</strong> keep xmlrpc.php open, but secure your site properly.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=aceecb09ab7b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[5 Reasons WooCommerce Is Slow (and How to Fix It)]]></title>
            <link>https://medium.com/@kubelet/5-reasons-woocommerce-is-slow-and-how-to-fix-it-9c18bd893399?source=rss-144945e1e05d------2</link>
            <guid isPermaLink="false">https://medium.com/p/9c18bd893399</guid>
            <category><![CDATA[web-hosting]]></category>
            <category><![CDATA[wordpress]]></category>
            <category><![CDATA[ecommerce]]></category>
            <category><![CDATA[web-development]]></category>
            <category><![CDATA[woocommerce]]></category>
            <dc:creator><![CDATA[Karl Kubelet]]></dc:creator>
            <pubDate>Fri, 18 Jul 2025 09:58:18 GMT</pubDate>
            <atom:updated>2026-06-03T20:13:48.329Z</atom:updated>
            <content:encoded><![CDATA[<p>WooCommerce has a reputation for being slow. Most performance problems come down to misconfiguration, bad hosting, and bloated plugins. Here are five common reasons your WooCommerce store is underperforming, and what you can do to fix it.</p><p>👋 <em>Hey there, I’m Karl, and I’m currently working on </em><a href="https://koddrio.com/"><em>Koddrio Cloud</em></a><em>, a WordPress hosting platform for agencies and power users. Please reach out if you’d like to give it spin! Now, back to the post.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6xv97zdGFX9odtvV-O8DeA.png" /></figure><h3>1. You Are Not Caching Enough</h3><p>A lot of store owners think that e-commerce must be fully dynamic, and therefore caching is off the table. That is wrong. Most of your visitors are not adding products to their cart. They are browsing. You can and should cache those requests.</p><p>If your cache hit rate is below 90 percent, you are leaving performance on the table. A properly configured caching layer should only bypass cache when a user adds something to their cart, logs in, or hits the checkout. Everything else should be lightning fast.</p><h3>2. Misconfigured Caching Because of Useless Cookies</h3><p>The number of WooCommerce sites running full-page caching with a near-zero hit rate is staggering. The culprit is often marketing and analytics plugins that write cookies on every single request. These are usually JS-only tools that do not need to interfere with caching.</p><p>If your cache is bypassing because of _fbp, _ga, _gid, or similar cookies, pixels and third-party analytics and ad services, fix your configuration. Either remove the plugin or add the cookie to your ignore list. Cloudflare, Nginx, Varnish, or whatever you use can all be told to ignore specific cookies. Do it.</p><h3>3. Bad Plugins That Destroy Your Database</h3><p>WooCommerce’s database structure is already a mess, something it inherited from Jigoshop, and built a bigger mess on top of over the years. Add a few poorly written plugins and you are asking for trouble. Features like faceted search, product recommendations, and dynamic filtering are all easy to implement the wrong way.</p><p>Many of these plugins run absurd SQL queries, sometimes joining a dozen tables, using wildcards, or scanning millions of rows. Things might look okay when you have ten products, but scale to a thousand and your site starts crawling.</p><p>If your store is slow and you have dozens of Woo plugins installed, start trimming. Better yet, test the performance impact of each one.</p><h3>4. You Are Not Benchmarking the Right Things</h3><p>Performance tuning without benchmarks is guesswork. You need to test critical paths like:</p><ul><li>Viewing a product page</li><li>Adding to cart</li><li>Loading the cart</li><li>Checking out</li></ul><p>Start with a fresh WooCommerce install, no plugins, default theme. That is your baseline. Now test your actual store. If a simple action takes 200 milliseconds on baseline and 2 seconds on your site, you know where to look.</p><p>Use tools like <a href="https://xdebug.org/">Xdebug</a>, <a href="https://wordpress.org/plugins/query-monitor/">Query Monitor</a>, <a href="https://www.php.net/manual/en/book.xhprof.php">XHProf</a>, or even just logging response times. You will find the outliers quickly.</p><h3>5. You Are Running on Terrible Hardware</h3><p>This is the real kicker. WooCommerce is CPU-bound.</p><p>Yes, you can have 17 analytics plugins doing stupid cURL requests to external slow services, or weird database queries with LIKEs on meta_value, or try reading 400 individual keys from Redis on each request, which certainly makes things IO-bound.</p><p>But once you eliminate those stupid outliers, you’re CPU-bound, and this is the baseline. Yet most hosting providers give you garbage compute, then try to upsell you to VPS plans with more garbage.</p><p>I saw a plan recently offering 300 PHP workers on a 6-core VPS. What kind of cores are those? That is not how concurrency works. Don’t fall for that nonsense.</p><p>We ran benchmarks on high-end managed Woo hosting and top-tier cloud providers. Then we ran the same tests on a 16-core dedicated server for the same price. The dedicated server crushed everyone. Not by a little. Ten times better performance than the fastest managed host. It wasn’t even close!</p><p>If you are serious about WooCommerce, you need serious hardware. That means bare metal, high core count, tuned PHP and MySQL. The cloud will not save you. Convenient “CPU-optimized” VMs will drain your pockets with nothing to show for.</p><p>P.S. If you want someone to tune and manage your WooCommerce server properly, <a href="https://koddr.io/hosting.html">get in touch</a>. I do this every day. And we know exactly where Woo falls apart.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=9c18bd893399" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>