A sanitized log from a day spent putting ingress in front of the same service two different ways, disproving a "CLI bug" that turned out to be a stale artifact, and wiring open-source analytics into a static site. The specifics are private; the patterns are the point.
A sanitized log from a day split between site automation, cloud-provider edge cases, certificate tooling, and a noisy gateway investigation. The details are private; the useful parts are the patterns.
An experiment: a daily, public log of what I worked on — stripped down to the parts that are actually transferable. No customer names, no internal ticket numbers, just the engineering lessons that survive being taken out of context.
Today had three things worth writing down.
Another sanitized log — the transferable parts of a day that was mostly "where exactly does this run, and is the thing I think is fixed actually fixed?". Four lessons worth keeping.
When you see too many open files in a containerized app, it’s tempting to jump straight to ulimit -n. Sometimes that’s correct. But on Linux (especially with Go apps using fsnotify), the error can also be caused by inotify limits—even if your process has a huge file-descriptor limit.
This post is a practical, copy/paste-friendly checklist to debug the problem on a real Kubernetes cluster.
Have you ever had a service that works perfectly on localhost but acts like it doesn't exist to the rest of your network? We recently spent an afternoon debugging a Mac Mini M4 that refused to accept SSH connections, despite every local check saying "All Systems Go."
The culprit wasn't a firewall or a wrong config, it was the invisible wall between 5GHz and 6GHz WiFi bands.
This tutorial is a guide on how to deploy an application in Kubernetes that will authenticate using Azure Workload Identity on Azure Kubernetes Services (AKS).
This tutorial contains a working example of exposing TCP services (LDAP/LDAPS + SSH) from a single-node k3s cluster running on an EC2 instance, using:
TCPRouteHTTPRoute/GRPCRoute (example file included)The key implementation detail for k3s-on-EC2 with the default overlay networking (flannel): use instance targets + NodePorts for L4 routes. ClusterIP + pod IP targets won’t work unless pods are VPC-routable (AWS VPC CNI).
A lot of our work nowadays requires using and connecting to a Virtual Private Networks (VPNs) in order to access certain resources (e.g. databases, websites, REST APIs) that were deemed important to protect from the public internet. When we connect to the VPN, we're able to access these resources.
The nature of our modern digital work requires simultaneous access to a plethora of services. Some of these services require an active VPN connection and some can be accessed without.
At times, the VPN we need to connect to is geolocated far from us. In addition, the VPN can be one that serves the entire company and is not very performant. These factors result in an experience of collective latency accessing resources, restricted and unrestricted ones alike.
If you find/found yourself in this type of situation before, this post will explain how you can circumvent that and suffer latency only when accessing the restricted resources instead of all resources by modifying the operating system routing tables in what's called 'Split Tunneling'.
To do this you will need to have root/administrator access to the UNIX operating system.
We begin by collecting the relevant information and later performing the modifications.
The company I work for decided a few months ago that we’ll be moving all customer tickets and assets from two separate systems (one for chat and one for old-school tickets) into a new, integrated system which provides both capabilites. My task was to perform the migration between the systems. Even though I’m not data engineer by any means, I accepted the challenge and thought it would teach me a lot about the planning and execution of such a complex project. It would also allow me to hone in my development/scripting skills and finally have some hands-on experience using a Python library I was always interested in working with, pandas.