Docker on Luis Sousa Blog

Fixing 'Robots.txt Unreachable' in Google Search Console with Traefik and Docker

Sun, 08 Mar 2026 10:00:00 +0000

We run a travel blog (joyofexploringtheworld.com) on Docker Compose with Traefik v3 and Cloudflare. One morning Google Search Console showed every page blocked from indexing with “Failed: Robots.txt unreachable”. The site was working fine in a browser, so what was going on?

Two separate issues were conspiring to break Googlebot’s ability to fetch /robots.txt. Here’s what we found and how we fixed both.

The setup Link to heading

Our WordPress service sits on two Docker networks: app-network (shared with Traefik, Redis, imgproxy) and db-network (shared with MariaDB). We run two scaled WordPress containers behind Traefik’s load balancer.

Why Datadog Apache Metrics Showed 404 (And When to Disable the Check)

Sat, 07 Mar 2026 12:12:00 +0000

We run joyofexploringtheworld.com with Datadog for observability. After setting up the Datadog Agent, we noticed the Apache integration was logging 404 errors every 15 seconds. Here’s why—and two ways to fix it.

The problem Link to heading

Datadog’s Apache integration polls http://localhost/server-status?auto to scrape metrics (requests per second, busy workers, etc.). The official wordpress:latest Docker image does not enable Apache’s mod_status module. So the endpoint doesn’t exist, and every poll returns 404.

Automated WordPress Backups to Hetzner with Docker (Zero Plugins)

Sat, 07 Mar 2026 12:08:00 +0000

We run a travel blog at joyofexploringtheworld.com on a single VPS with Docker Compose. Managed backup plugins add overhead, cost money, and only back up what WordPress can see. We wanted to back up everything—volumes, configs, database—automatically, with offsite copies. Here’s how we did it with zero WordPress plugins.

The backup service Link to heading

We use offen/docker-volume-backup as a sidecar container in our Docker Compose stack. It runs on a cron schedule, tars up the mounted volumes, and ships the archive to a remote Hetzner Storage Box via SFTP.

Self-hosted image optimization for WordPress with imgproxy (no premium plugins)

Sat, 07 Mar 2026 12:06:00 +0000

We run a travel blog (joyofexploringtheworld.com) on a budget VPS with Docker Compose and free Cloudflare. Premium image plugins like ShortPixel charge for resize and optimize—we wanted dynamic images without recurring costs. Here’s how we did it with imgproxy.

Why imgproxy Link to heading

imgproxy runs in Docker, supports signed URLs and allowlisted sources, and outputs WebP/AVIF. We use the official image ghcr.io/imgproxy/imgproxy and pin a version tag for stable deploys. No post edits needed: we rewrite image URLs at runtime via a WordPress MU-plugin so existing content and our Bricksy Pro FSE output stay unchanged.

Running WordPress Cron the Right Way When Your Site Lives in Docker

Sat, 07 Mar 2026 12:05:00 +0000

We host a travel blog (joyofexploringtheworld.com) on a low-cost VPS with Docker Compose and Cloudflare in front. WordPress’s built-in cron was causing intermittent timeouts in our logs. Here’s how we fixed it.

The problem with WordPress cron in Docker Link to heading

WordPress cron normally “pings itself” on each page load—an HTTP POST to the site from inside the container. Behind a CDN and in Docker, that ping can time out, fail, or hit the wrong instance. We saw cURL timeouts in Datadog and unreliable scheduled tasks.

How We Sped Up Our Travel Blog (Without Changing Hosts)

Sat, 07 Mar 2026 12:00:00 +0000

We run a travel blog at joyofexploringtheworld.com that helps travellers plan itineraries—hosted on a low-cost VPS with Docker Compose and Cloudflare’s free tier. Here are four changes that improved speed without changing hosts or paying for premium plugins.

1. Give the database more memory Link to heading

On a 24 GB VPS, MariaDB’s default InnoDB buffer pool (~128 MB) was too small. We increased it to 2 GB and set a 4 GB memory limit on the DB container. Redis’s maxmemory was raised from 1 GB to 2 GB.

How We Speeded Up Our Travel Blog (Without Changing Hosts)

Sat, 07 Mar 2026 12:00:00 +0000

1. Give the database more memory Link to heading

Why Our Site Went Down for an Hour (And What We Fixed)

Sat, 07 Mar 2026 10:00:00 +0000

We run a travel blog (joyofexploringtheworld.com) that helps people plan itineraries—hosted on a low-cost VPS with Docker Compose and Cloudflare. One small performance tweak took the whole site offline for an hour. Here’s what happened and how we fixed it.

What Went Wrong Link to heading

We added a Traefik compression middleware to speed up responses. The new config lived in a separate YAML file. Traefik was configured to load only a single file (e.g. real-ip.yaml), not the whole directory. The router referenced compress@file, but that file was never loaded. Every request failed.

Running a WordPress Travel Blog on a Budget VPS: The Full Docker Stack

Sat, 07 Mar 2026 09:00:00 +0000

We run joyofexploringtheworld.com — a travel blog — on a single 24 GB VPS using Docker Compose and free-tier Cloudflare. No managed WordPress hosting, no premium plugins, no surprise invoices. Here is the full stack, laid out so you can steal whatever is useful.

Architecture overview Link to heading

The request path looks like this:

Internet
  -> Cloudflare (CDN / APO / edge SSL)
    -> Traefik v3 (reverse proxy, TLS termination for origin)
      -> 2 WordPress containers (round-robin with sticky session cookies)
        -> MariaDB 11 (single-writer database)
        -> Redis (persistent object cache)

Alongside the main request path we run a handful of supporting services: imgproxy on a dedicated subdomain for on-the-fly image resizing and format conversion, a wp-cron sidecar that hits wp-cron.php every five minutes so we can disable the default front-end cron, an automated backup container that dumps the database and syncs uploads to Hetzner Storage Box nightly, and the Datadog agent for logs, traces, and container metrics.