Home on Igor Kupczyński

YOLO Mode in a Vagrant Box

Mon, 02 Feb 2026 12:00:00 +0100

A Vagrant-based sandbox for running AI coding agents in full autonomous mode—without risking your host machine.

The problem #

AI coding agents work best when they can run autonomously. Claude Code’s claude --dangerously-skip-permissions and Codex’s codex --full-auto let agents execute commands without asking for approval at each step. But this is scary for good reason—agents have deleted home directories, wiped working directories, and destroyed files while misreporting what happened.

Solution: disposable VMs #

People use devcontainers, Docker, or remote VMs to isolate their agents. I wanted something I can blow away and recreate in seconds. Vagrant is a bit of an overkill, but it works.

Pattern Spotting: Hatchet

Tue, 13 Jan 2026 00:00:00 +0000

Hatchet is a durable task orchestration platform, built on PostgreSQL. Competitors include Celery or Temporal. Hatchet scaled from 20k tasks/month to 1B+/month without abandoning Postgres.

Some patterns caught my attention.

PostgreSQL as the Universal Backend #

Conventional wisdom says separate your queue (Redis/RabbitMQ) from your state (PostgreSQL). Different access patterns, different tools. Hatchet bets against this using Postgres for everything.

The queue implementation relies on FOR UPDATE SKIP LOCKED, a Postgres feature that lets workers claim rows without blocking each other.

GCP Live-Migrates, AWS Reboots: How Cloud Providers Handle Host Maintenance

Tue, 13 Jan 2026 12:00:00 +0100

The following incident is fictional, but the architecture lessons are real.

The incident #

Monday morning. You’re reviewing last week’s metrics and notice a gap in the graphs — your staging environment was down for about 5 minutes on Thursday afternoon. The AWS console shows the instance is healthy, but uptime is only 4 days.

Digging into the EC2 console, you find it: AWS had sent a Scheduled Event notification two weeks prior. The instance was due for host maintenance. The notification sat in the console unread. The instance rebooted, dropping connections and clearing the in-memory cache.

TCP connection limiting in Go: Blocking Accept vs Active Load Shedding

Wed, 19 Nov 2025 00:00:00 +0000

I recently reviewed a Go codebase that implemented connection limiting in a TCP proxy. The goal was to protect the backend from being overwhelmed by too many concurrent connections.

The initial implementation used a semaphore to gatekeep the Accept() call. While this seems intuitive, it relies on the kernel’s backlog behavior in ways that can cause significant operational issues.

TL;DR #

Don’t block before calling Accept(). If you do, the OS continues to complete TCP handshakes and queues connections in the kernel backlog. Clients believe they are connected, but your application is not processing them.

AGENTS.md: A Standard for AI Coding Agents

Sun, 05 Oct 2025 00:00:00 +0000

If you’re using AI coding agents—and who isn’t these days—you’ve probably noticed each tool wants its own configuration file. Cursor reads .cursor/rules, Claude looks for CLAUDE.md, and every new agent seems to invent yet another convention. This proliferation of tool-specific files is annoying.

Enter AGENTS.md: a unified standard that consolidates project context for AI agents in one predictable location.

The Problem #

Modern development environments are heterogeneous. You might use Cursor for rapid prototyping, your buddy uses Claude for architectural discussions, and open-source contributors prefer Aider or GitHub Copilot. Each agent needs to understand your project structure, build commands, coding conventions, and architectural decisions. Maintaining separate configuration files for each tool means duplicating content and inevitably letting some documentation drift out of sync.

Why AI Agents Hit the Wall

Tue, 29 Jul 2025 00:00:00 +0000

I found this excellent article by Utkarsh Kanwat: Betting Against Agents.

Compounding Works Both Ways #

We usually talk about compounding in positive terms—interest accumulates, wealth grows, skills build on skills. But with AI agents, compounding works against you: the failure rate of an agentic process compounds with each step.

This framing really clicked for me. I’ve noticed that I need to break problems into small steps and have agents implement incrementally, verifying each step. The article gives intuition into why: even with 95% reliability per step, you’re down to 36% success rate after just 20 steps. Not good enough for production systems.

Effective Prompt Engineering in 2025

Fri, 16 May 2025 00:00:00 +0000

I recently gave a talk at an internal Elastic conference about prompt engineering techniques in 2025, accompanied by live demonstrations using Mistral via Ollama, VS Code Copilot, and Windsurf. Here’s a brief overview of the key topics covered.

Why Prompt Engineering Matters #

Prompt engineering dramatically improves output quality, enables structured responses, and allows the use of smaller, more cost-effective models. The economic impact is substantial:

Model	Input (per 1M tokens)	Output (per 1M tokens)
GPT-4.1	$2.00	$8.00
Mistral Small 3	$0.10	$0.30

Core Concepts Covered #

My presentation explored several key areas:

Raycast browser switcher: a vibe coding rabbit hole

Sat, 03 May 2025 00:00:00 +0000

I often change my default browser. I use Chrome for work and Safari for personal.

It would be nice to have this in Raycast. Raycast is a Spotlight-like app launcher for macOS. Ideal UX: type chbr, get a list of browsers, pick the one you want to be the default, and hit return.

Sounds simple enough. The problem? I’m neither a TypeScript nor a macOS developer.

Vibe coding to the rescue #

In the age of LLMs this shouldn’t be a problem. I can vibe-code the solution¹.

Principal Engineer Roles Framework

Wed, 22 Jan 2025 00:00:00 +0000

I came across an interesting article by Mai-Lan Tomsen Bukovec, AWS Vice President of Technology, about a framework for Principal Engineer roles. The framework originated from their observations of how Principal Engineers work at AWS, particularly in the S3 team. I’m summarizing it here for easier reference.

The Six Principal Engineer Roles #

The framework defines six distinct roles that Principal Engineers can play:

A project/program lead spanning multiple teams. They drive project success by ensuring decisions are made and removing obstacles. Sponsors own all aspects of project delivery, including technical direction, product definition, and organizational alignment. They need to:

DNS failover: When zero-weight records strike back

Wed, 27 Nov 2024 00:00:00 +0000

Recently, I encountered a fascinating incident where some API endpoints received unexpected traffic. Services were getting requests they weren’t configured to handle, which shouldn’t have been possible with our ALB routing rules. Here’s what I learned.

Details of this incident and the service involved are anonymized to not distract from the key learning.

TL;DR #

When all weighted DNS records with weight > 0 become unhealthy, Route53 falls back to considering zero-weighted records. This can unexpectedly activate legacy request flow with different routing behaviors. Always fully decommission legacy infrastructure rather than leaving it in DNS with zero weights.

Chain of Density: An LLM summarization technique

Tue, 17 Sep 2024 00:00:00 +0000

I came across an interesting method called “Chain of Density” that promises to generate increasingly concise and information-rich summaries. Let me share what I’ve learned.

Chain of Density is an iterative summarization technique that aims to create progressively denser summaries of a given text. The process involves repeating two key steps multiple times:

Identify 1-3 informative entities from the original text that are missing from the previous summary.
Write a new summary of identical length that incorporates all the information from the previous summary plus the newly identified entities.

What makes this method particularly interesting is its focus on maintaining a consistent word count while increasing information density. This constraint forces the summarization process to become more efficient with each iteration, resulting in highly compact yet comprehensive summaries.

Founder Mode

Mon, 02 Sep 2024 00:00:00 +0000

Paul Graham’s “Founder Mode”

Paul Graham’s essay Founder Mode is doing the rounds in social media:

In effect there are two different ways to run a company: founder mode and manager mode. Till now most people even in Silicon Valley have implicitly assumed that scaling a startup meant switching to manager mode.

There are as far as I know no books specifically about founder mode. Business schools don’t know it exists. All we have so far are the experiments of individual founders who’ve been figuring it out for themselves.

Amazon S3 now supports conditional writes

Tue, 27 Aug 2024 00:00:00 +0000

TIL S3 supports conditional writes

Check out the official AWS announcement and the more detailed S3 Conditional requests guide.

Amazon S3 adds support for conditional writes that can check for the existence of an object before creating it. (…) You can perform conditional writes using PutObject or CompleteMultipartUpload API requests in both general purpose and directory buckets.

…

To use conditional writes, you can add the HTTP if-none-match conditional header along with PutObject and CompleteMultipartUpload API requests.

Amp It Up

Sat, 11 May 2024 00:00:00 +0000

Great book by the star CEO of Snowflake.

Amp It Up by Frank Slootman was highly recommended on ~~Twitter~~ X and Hacker News.

Frank Slootman was the CEO of a couple of high-profile companies that he took to IPO: Snowflake, Service Now, and Data Domain. In the book, he shares his management philosophy and supporting stories.

The main pillars are:

Raise your standards
Align your people and culture
Sharpen your focus
Pick up the pace
Transform your strategy

⠀ Common motifs that run through the stories are:

Make Go application container resource limits aware

Thu, 11 Jan 2024 00:00:00 +0000

Go binaries are not inherently container-aware, specifically in that they do not account for memory and CPU limits.

TL;DR

When running a Go application in a container with set CPU or memory limits, you should inform the Go runtime of these limits using the GOMAXPROCS and GOMEMLIMIT environment variables.

For example, if you run your application as follows:

docker run --cpus="4.0" --memory="2000m" my-go-app

Inform the Go runtime as follows:

docker run --cpus="4.0" --memory="2000m" -e GOMAXPROCS=4 -e GOMEMLIMIT=1800m my-go-app

Memory

Docker and Kubernetes use cgroups to enforce memory limits, which means that if a process requests more memory from the OS than its assigned limit, it may be terminated by the oom-killer (which is the out-of-memory subsystem of the Linux kernel):

Golang 1.20 upgrade failed due to GLIBC_2.34 not found

Mon, 07 Aug 2023 00:00:00 +0000

Remember that Java slogan, Write once, run anywhere? Go compiles directly to binaries, which target specific architectures. There’s no intermediate byte code. You cross-compile, and once you produce a binary for the target architecture: GOOS=linux GOARCH=amd64 go build, it just works. Well, at least usually.

With Cgo, go can interop with C code. There’s some magic involved, the toolchain behaves differently depending on your path (see below), but you usually don’t need to worry about it at all.

Negotiate your salary

Sun, 01 May 2022 00:00:00 +0000

Bookmarks on salary negotiation:

Classic from Patrick McKenzie
- Salary Negotiation: Make More Money, Be More Valued
101 on offer negotiation from Rishi Taparia
- Negotiating your startup job offer
Other links in no particular order

Books:

My favorite is Never Split the Difference by Chris Voss
Timeless classic How to Win Friends and Influence People by Dale Carnegie

Private Link is the IP filtering of the cloud

Sun, 30 Jan 2022 00:00:00 +0000

Use cases for Private Link and differences in its implementation across the major Cloud Providers.

Changelog

November 27, 2024: Updated to reflect AWS PrivateLink’s new cross-region connectivity support. This feature brings AWS PrivateLink’s capabilities more in line with Azure and GCP’s offerings.
January 30, 2022: Original post published.

Private Link #

Private Link allows secure connectivity to services hosted on different Cloud provider accounts.

Private Link allows you to connect to a service in a different VPC—and that VPC can be owned by a different account.
The connection is secure¹:
- It doesn’t leave the cloud provider’s network.
- Data exchanged over Private Link is encrypted (I wasn’t able to find much on the encryption in the documentation—it makes sense to run secure the connection with TLS).
- You can use native features, such as IAM and security policies, to limit who can access the connection.
The connection is unidirectional.
- Resources in the source VPC can access a dedicated service in the target VPC.
- Other resources in the target VPC are not exposed.
- Resources in the source VPC are not exposed.

The Phoenix Project

Fri, 30 Apr 2021 00:00:00 +0000

Theory of constraints applied to IT operations.

I’ve read The Phoenix Project: A Novel about IT, DevOps, and Helping Your Business Win by Gene Kim, Kevin Behr, George Spafford. This book is amazing. If you follow the lean manufacturing practices, or enjoyed The Goal (see The Slowest Hiker), or simply want to improve the way IT operations work with the rest of the org (and improve the org in the process!) you’ll enjoy the book.

Fun with docker networks

Mon, 22 Mar 2021 00:00:00 +0000

How to find out a container responsible for a given open port on a host.

Recently, I’ve debugged an issue with a server running multiple docker containers. They interact with each other by exposing TCP ports on the host and communicating over these ports.

In this post:

I describe a surprising discovery about host network mode containers (spoiler: they don’t show under PORTS in docker ps),
Show to find a container responsible for a given port on the host,
Link to a repo where I investigate docker networking in much more detail.

The story #

Let’s say we have the container exposing :32080 on the host:

Notes on creating a telegram bot

Sat, 28 Nov 2020 00:00:00 +0000

I’ve created a telegram bot to give me a daily updates to a webpage I was tracking. This is a hobby project and the hosting price is an important concern.

The problem
- Context
The solution
- Why not open-source the code?
Notes on the project
Summary

The problem #

I want to be notified about new real estate ads from my neighborhood.

Remote debug your go code

Sun, 17 May 2020 00:00:00 +0000

This post has a companion github repo github.com/igor-kupczynski/remote-debug-example – check the repo for a simple example.

Coming to go from the JVM background I often find myself looking for analogs of the features I rely on the JVM ecosystem. One example of such a feature is remote debugging – which is built into JVMs. Read on to find out how to accomplish this in go.

Remote debug #

Remote debug is a very useful feature. You start your application with special options and then you can connect to it with a debugger from your IDE. This is very useful if you want to debug an environmental issue (e.g. that manifests itself only on Google Cloud Platform, or only in QA, but you can’t easily reproduce on your laptop). Maybe even more common nowadays – you may want to debug an application which is running in a docker container.

Natural project planning

Sat, 08 Feb 2020 00:00:00 +0000

Natural project planning is a project planning method described by David Allen in his bestseller Getting Things Done. In this post, I describe the method.

Natural project planning consists of a couple of steps:

Define purpose and principles.
Visualize goals and outcomes.
Brainstorm.
Organize the ideas.
Define next actions.

What are the goals of natural project planning?

Help clarify the steps needed to complete the project.
Visualize the endgame — the outcome of the project.
Identify next actions needed to move the project forward, and people responsible to complete them.

This is not a strict process and you should not follow it religiously. Follow it only as much as it makes you feel you have the project under control. You can always redo any of the steps later — or zoom in as needed.

FIPS compliant crypto in golang

Sun, 15 Dec 2019 00:00:00 +0000

Updated: 2022-09-29. To reflect that BoringCrypto is now part of the main branch.

Updated: 2022-11-20. To add a section on TLS 1.3 and FIPS.

Updated: 2023-01-07. To add a section on automatic configuration at compile time: Force FIPS-compliant crypto at compile time.

Updated: 2023-08-07. To add a note of microsoft/go.

If you work with US government entities or corporations in regulated markets the subject of FIPS compliance may come up. FIPS 140-2 is a set of cryptographic standards that your application may need to adhere to.

Zero to One

Tue, 08 Oct 2019 00:00:00 +0000

I’ve read Zero to One: Notes on Startups, or How to Build the Future by Peter Thiel. Here are my notes on the book.

Background #

Peter Thiel is a well known Silicon Valley founder and investor. He is often portrait as a contrarian and is quite a controversial figure. He is a founder of PayPal and Plantir. I’ve heard very good reviews of the book and it was on my list for quite some time.

Elastic Management Philosophy

Sat, 25 May 2019 00:00:00 +0000

How to build a great engineering organization — advice from Elastic’s SVP Engineering.

Update 2019-11-29: Kevin published Leadership @ Elastic blogpost. Kevin’s post is more of high-level and strategical than the content here.

I’ve recently come back from Elastic’s Global All Hands 2019 (or simply GAH 2019). It is a conference-style internal Elastic event. The goal is to facilitate cross-team communication, better understand the company strategy, and to simply chat with your colleagues. This post contains notes from Kevin Kluge’s Management Philosophy session.

Server Name Indication with Traefik and Go

Tue, 21 May 2019 00:00:00 +0000

How to setup Server Name Indication with Traefik.

In this blog post we discuss briefly what an SNI is, show a simple example of setting up SNI with Traefik — a popular reverse proxy written in go — and then show some options for implementing it in “plain” go. This blog has a companion repo traefik-tls-demo, be sure to check it if you want to try the example.

Server Name Indication (SNI) #

To establish an https connection the server presents its certificate to clients as part of the TLS handshake. Since the connection is not established yet, the server doesn’t see the http request — neither the request line, not its headers. We’re not at the application level yet. Therefore, the server has to make the decision on which certificate to present based on the target IP of the connection. In practice, this limits the server to a single certificate per public IP address.

The Slowest Hiker

Sat, 23 Feb 2019 00:00:00 +0000

How to improve any process.

One of the two most often requested features of Go is the support for generics. Arguably, generics are super important and they are taken for granted in most of the modern programming languages. In fact, Go developers get a lot of flak from their Haskell-minded colleagues being forced to use a language with such a poor feature set. And yet, Go is one of the fastest growing languages on Github. It seems like it became the lingua franca of the cloud, with more and more distributed systems being written (or rewritten) in Go.

Set up AWS Lambda with terraform

Sat, 19 Jan 2019 00:00:00 +0000

In the previous post, we’ve presented an AWS Lambda function to automatically invalidate resources in CloudFront distribution when underlying objects in an S3 bucket change.

Instead of clicking and copy-pasting in the AWS Console, we can use terraform to set this function up. In fact, I’ve already made it a part of my terraform static aws website — terraform module which sets up an S3 bucket to host a static website and CloudFront as a cache; it also handles a redirect www.domain.com --> domain.com and, provided with an AWS generated https cert, the https:// bit.

AWS Lambda to invalidate CloudFront when S3 object changes

Wed, 09 Jan 2019 00:00:00 +0000

There are only two hard things in Computer Science: cache invalidation and naming things.

– Phil Karlton

In this post we will focus on cache invalidation.

Problem statement #

This blog is a static site hosted on S3 with CloudFront being used as a CDN / cache. Blog post about this setup.

UserAgent requests kupczynski.info/posts/abc.html from CloudFront.
If this is not in the CloudFront cache, then we have a miss — CloudFront consults S3 bucket for an up-to-date version.
On next requests to this resource it is in the CloudFront cache, so we have a hit. No need to consult S3 bucket.
Occasionally, a post is updated by my Continuous Integration infrastructure.

The problem is how to serve an up-to-date version of a post. And make sure that we hit the cache most of the time.

OpenVPN and DNSes on Ubuntu

Wed, 09 Jan 2019 00:00:00 +0000

One of the VPNs I connect to sets the DNS server for the link. It is an OpenVPN with DNS option. The problem is that is doesn’t really work out of the box in Ubuntu (at least 18.04 and 18.10) — the VPN DNS is not consulted.

At the time of writing, I’m on 18.10, and I write it from that perspective. Ubuntu uses systemd-resolved service to provide a local DNS resolution. We need to “push” the VPN DNS address to this service. We can do it using jonathanio/update-systemd-resolved helper script.

Fire and Motion

Sat, 08 Dec 2018 00:00:00 +0000

Today I’ve rediscovered Fire and Motion blogpost by Joel Spolsky. I recommend you to read it.

It start with how you can get only a couple of productive hours per day. And this is OK. Joel mentions a colleague, who worked only from 12 to 5, and still had more done than the average other team member. These productive hours are called the flow.

One of the tricks to put yourself in the flow is to start. Starting is often the hardest thing. This is way better understood now, than in 2002. With a lot of articles and books on the importance of deep work, the cost of multi-tasking, etc.

Change the maximum number of open files in ubuntu 18.10

Sat, 24 Nov 2018 00:00:00 +0000

The procedure changed on recent ubuntu compared to what I used to do. I used change /etc/security/limits.conf.

The procedure on ubuntu 18.10 that worked for me is this.

Go to /etc/systemd/system.conf

Uncomment DefaultLimitNOFILE and set your limit there, e.g.

$ grep NOFILE /etc/systemd/system.conf
DefaultLimitNOFILE=65535

Restart
Profit

Before:

$ ulimit -n
1024
$ ulimit -Sn
1024
$ ulimit -Hn
1048576

After:

$ ulimit -n
65535
$ ulimit -Sn
65535
$ ulimit -Hn
65535

The limits can be controlled by systemd and this is what we do here — instruct systemd to set it to 65k. Note that this setting will apply to all users, which is OK for a laptop or a workstation. On a server you may need a finer grained settings and you should look for a different method.

Nvidia on Ubuntu 18.10 requires lightdm

Sun, 18 Nov 2018 00:00:00 +0000

After a upgrade from 18.04 my external monitor stopped working. Installing lightdm was the solution.

Update 2020-11-10. Abish Vijayan posted a solution for Ubuntu 20.10 in the comments below.

Made it to work. Disabled Wayland in gdm3. https://askubuntu.com/questions/975094/how-to-disable-wayland-in-17-10-in-gdm3-login-screen

Update 2019-11-15 on Ubuntu 19.10 I still need the lightdm to support multiple monitors.

Problem #

I’ve recently updated Ubuntu from 18.04 to 18.10 on work-issued Thinkpad p51. The laptop has nvidia quatro m2200 graphics it also has a 4k display and I use an external 4k monitor most of the time.

Use wine to run windows JDK on linux

Sun, 11 Nov 2018 00:00:00 +0000

I’ve got a legacy windows app that I need to run from time to time. The app is written in java, but uses some 32 bit JNI dlls. I’ve used to keep a windows VM around to run it, but recently I was able to “port” it to wine.

In short, you need install wine on ubuntu:

# I've needed a 32 bit version
$ sudo dpkg --add-architecture i386
$ sudo apt-get -y update && sudo apt-get install wine32

Then get the 32bit JDK from Oracle. It needs to be JDK, JRE wasn’t working for me (and some folks on the internet). As of time of writing I’ve got JDK 8u192 and this is what I’ve tested with.

Diagnose SSD failures

Fri, 02 Nov 2018 00:00:00 +0000

How to minitor health of SSD drives, case study on my failed Kingston SUV500MS480G.

I use my old Samsung Series 7 ultrabook for some personal tasks. It originally had 120 GB SSD, but I’ve upgraded it recently to Kingston 480GB SSD.

I’ve first installed Ubuntu 18.04, but then upgraded to 18.10. Everything seem fine, but after few hours of usage and a restart it started complaining about invalid file permissions. I don’t have logs, as at this point I though — ok, a botched upgrade — we can reinstall, no big deal. Reinstalling is not a big deal, because most of the data is in the cloud already, and I maintain a collection of scripts to setup the software I need.

Ubuntu 18.04 on Thinkpad P51

Sun, 30 Sep 2018 00:00:00 +0000

Back to Linux after almost 3 year hiatus.

Switching back from MacBook to Thinkapad and Ubuntu as the main work machine.

My MBP 13’’ aged quite nicely, but compiling large scala codebases and running VMs / containers became quite frustrating. I was eligible for a refresher at Elastic and I’ve decided to try Thinkpad P51 + Ubuntu. This is a 15’’ beast with 64 GB of RAM. It is quite heavy, but working remotely I don’t need to carry it to the office every day, while still being able to travel if needed. I consider the trade off acceptable.

Zoom and Kubuntu 18.04

Sat, 08 Sep 2018 00:00:00 +0000

Fix annoying issues with Zoom and HiDPI on Kubuntu 18.04.

I’ve got a new laptop for my work for Elastic recently. This time I’ve opted for a Thinkpad and not for a MacBook. I’ve installed Kubuntu 18.04. Linux on desktop or laptop gives you a lot of opportunities to blog about something you’ve fixed. Here comes a first one.

How to make zoom — the conferencing software — play nicely with kubntu?

JMH - Java Microbenchmark Harness

Thu, 12 Jul 2018 00:00:00 +0000

Notes I’ve took while doing a microbenchmark of some code. Not really structured, but maybe useful for future me.

Webpage: http://openjdk.java.net/projects/code-tools/jmh/

Getting started #

Use an archetype:

mvn archetype:generate \
 -DinteractiveMode=false \
 -DarchetypeGroupId=org.openjdk.jmh \
 -DarchetypeArtifactId=jmh-java-benchmark-archetype \
 -DgroupId=info.kupczynski \
 -DartifactId=jmh-playground \
 -Dversion=1.0

// Or for scala
mvn archetype:generate \
 -DinteractiveMode=false \
 -DarchetypeGroupId=org.openjdk.jmh \
 -DarchetypeArtifactId=jmh-scala-benchmark-archetype \
 -DgroupId=info.kupczynski \
 -DartifactId=jmh-scala-playground \
 -Dversion=1.0


mvn clean install

java -jar target/benchmarks.jar

Recommended approach:

setup a standalone project and depend on your jars

Some hints from the scala community https://github.com/ktoso/sbt-jmh

Who's Got the Monkey?

Wed, 27 Jun 2018 00:00:00 +0000

Classic advice on time management.

A colleague of mine (thanks Stu!) recently recommended me a classic article Management Time: Who’s Got the Monkey? It is classic indeed — published in 1974 . Some of the wording is in need of modernizing (e.g. the boss-subordinate hierarchy changed into much more flat organizations), but I think it aged quite well. I’ve enjoyed it and I recommend it to you, my reader :)

Here are my notes if you want to skim the article quickly.

JVM Class Data Sharing

Tue, 29 May 2018 00:00:00 +0000

This post describes how to enable Class Data Sharing (CDS) for a java app and what are the benefits of doing so.

I’ve recently seen a talk about CDS by Volker Simonis which was the inspiration to write this blogpost.

Table of Contents

What is Class Data Sharing?
Prepare CDS
- Intro — we need OpenJDK 10
Makefile
Experiment
- Let’s see how the classes are loaded
- Memory usage
Recap
Conclusions
- Ideas for future posts

Class Data Sharing is a JVM feature, which allows multiple JVMs to share loaded classes (and some other things) via shared memory.

Notes from Geecon 2018

Sun, 13 May 2018 00:00:00 +0000

Conference notes.

Geecon 2018 was an important event in Center- and Eastern- European JVM conference agenda. It was over a year since I’ve attended my last non-elastic related event so I thought it is a great venue to pick up some ideas and look where the industry is heading nowadays. And what a great venue it was!

In this pots I’ll present mostly raw notes I’ve took during the talks. If some of the ideas stick with me, I may explore and refine them later on. Oh, there where 4 concurrent talks at most of the slots, so I can cover only a subset. So the notes are uncomplete and may be misleading. Proceed at your own risk :)

Learning VIM with Spacemacs

Wed, 18 Apr 2018 00:00:00 +0000

I just don’t grok the vi (yet).

Three weeks ago, I’ve ditched most of my emacs config and installed spacemacs. The main driver was that I wanted to try modal editing instead of emacs key chords. I could’ve tried vim of course, but I’m already invested in emacs and feel at home there.

Evil-mode #

Luckily, there is a brilliant vim emulation in emacs, called evil mode. I didn’t want to spend too much time configuring it though plus I have no idea which vim plugins ported to emacs are worth looking at, so I’ve given spacemacs a go. Spacemacs is a distribution of emacs preconfigured with evil and with an extensive documentation.

12 Rules For Life

Mon, 02 Apr 2018 00:00:00 +0000

I’ve read 12 Rules for Life: An Antidote to Chaos by Jordan Peterson.

Background #

I’ve heard about Professor Jordan Peterson because of my colleague, who is a fan of him. I’ve watched the famous BBC interview with Cathy Newman. The interview itself is fun to watch as a lesson in debating. Feel free to google it, there are a plenty of recording and analysis of the interview on youtube. I’ve got intrigued and decided to give Peterson’s latest book a shot.

IT Career Options

Fri, 09 Mar 2018 00:00:00 +0000

Talk about career options for pupils of my old highschool

I was asked to do a short talk on the career options in IT for Henryk Sienkiewicz grammar school. I agreed without hesitation. It went quite well, I hope it helped some of the folks to made their mind.

It started with an interview (to be publish at schools website), and then a talk in a classroom format. For posterity, I’m attaching the slides below.

Shell in Emacs - Performance

Thu, 07 Dec 2017 00:00:00 +0000

Some of my coworkers use Eshell inside emacs as their main terminal emulator. I had a little bit of envy, so I decided to give it a try. Unfortunately, I’m not going to use it as my main shell. At least not yet.

How fast can you print? #

What is the issue then? — performance. I’ve run a build which outputs a lot of, mostly useless, warnings to the terminal. And boy, it took ages.

Insanely Simple: The Obsession That Drives Apple's Success

Fri, 20 Oct 2017 00:00:00 +0000

I’ve recently read Insanely Simple: The Obsession That Drives Apple Success by Ken Segall. This post summarizes my notes taken while reading the book.

Background #

Ken Segall is a marketing director who worked with Steve Jobs at Apple and NeXT for 10 years or so. He also worked with Dell, Intel, IBM and other tech companies. In the book he offers a lot of stories on how Apple and Steve Jobs (or Steve) did things. He contrasts it with the other tech companies. They are centered around few topics like think brutal or Think small. The book depicts Steve Jobs as a man devoted to simplicity and it suggests that simplicity is the key ingredient of the Apple’s success.

Elasticsearch and absolute_form in HTTP/1.1 request line

Fri, 08 Sep 2017 00:00:00 +0000

Elasticsearch doesn’t accept some of the HTTP/1.1 incompliant clients for a security reason.

Described on the elasticsearch forum. Elasticsearch complains when absolute_form, i.e. host + path, is present in the request line.

For example this request works (1):

POST /_search HTTP/1.1
host: localhost:9200
date: Fri, 28 Apr 2017 19:30:04 GMT
content-length: 42
content-type: application/json

{"query":{"match":{"_all":"Hello World"}}}

But this one doesn’t (2). Notice the http://localhost:9200 after POST:

POST http://localhost:9200/_search HTTP/1.1
host: localhost:9200
date: Fri, 28 Apr 2017 19:30:04 GMT
content-length: 42
content-type: application/json

{"query":{"match":{"_all":"Hello World"}}}

It fails with

Encrypt the traffic between nodes in your elasticsearch cluster

Sun, 02 Apr 2017 00:00:00 +0000

Encrypting the traffic between elasticsearch nodes using X-Pack security.

In my opinion, it is quite important to encrypt the traffic between the backend services. Especially, if you don’t control your infrastructure (or don’t trust your infra provider). Let’s see how to encrypt the elasticsearch cluster transport traffic with X-Pack.

X-Pack uses TLS to encrypt the traffic between nodes in the cluster and between clients and the cluster. Additionally, a hostname verification can be performed as well. The prerequisite for internal TLS is to assign a X.509 certificate to each node. The certificates should be sign by a trusted certificate authority, or CA.

How to put AWS Cloudfront in front of S3 hosting a static website

Mon, 06 Mar 2017 00:00:00 +0000

Some AWS API quirks make it hard to use folder level index documents via terraform.

If you use a static blogging software like jekyll or hugo you may be tempted to host them on S3. It is simple, cheap and has a great availability (lets ignore the recent us-east-1 incident for now).

The static blogging software relies on the fact that you can use index.html as a folder root resource. For example if you want to check the articles tagged as elasticsearch on this blog under the /tag/elasticsearch/ url you are in fact viewing /tag/elasticsearch/index.html generated by jekyll. Lucikly for us it is easy to set up S3 in a website mode which does just that.

Elastic Cloud Architecture

Wed, 25 May 2016 00:00:00 +0000

My talk at ElasticNL meetup in Nieuwegein.

I had a great opportunity to talk about Elastic Cloud architecture together with Morten Ingebrigtsen. This was an ElasticNL meetup. This time in a new venue at Nieuwegein.

The audience was great and we had a log of good questions. Thanks folks for attending.

You can grab the slides

Chamberconf 2016

Mon, 07 Mar 2016 00:00:00 +0000

The only fixed item on the schedule is the time of the diner!

I’ve just came back from Chamberconf 2016. It is quite an unusual conference on the Polish Java scene. It is small — around 80 people total. It is held in a palace, this year in Pałac Łagów. But most importantly, there are no sponsors and the form is not very strict.

The idea is to promote and facilitate informal discussions around the topics covered by speakers or any other JVM or CS related topcis. I think this year they met the expectations. There where some nice talks and tons of good discussions.

You Know, for Search...

Sun, 06 Mar 2016 00:00:00 +0000

Year 2016 brings new challenges. I joined Elastic to work on the Cloud team.

I’ve talked and blogged about elasticsearch quite a few times last year. At Egnyte, we’ve built a search engine which allowed us to offer great search experience to Egnyte’s customers. It was based elasticsearch and we’ve successfully scaled it up to billions of documents and dozens of terabytes of data. I had a lot of fun learning elasticsearch and working through proof-of-concept, implementation and then production deployment and scaling.

Elasticsearch in Production

Tue, 05 May 2015 00:00:00 +0000

Lessons Learned from Deploying and Managing Multi Billion Document Search Engine at Egnyte - my talk at Poznan Java User Group.

Elasticsearch in Production, Lessons Learned from Deploying and Managing Multi Billion Document Search Engine at Egnyte. I gave this talk on our search engine at Poznan Java User Group. You can find the slides here: https://kupczynski.info/t/2015-05-05-jug-elasticsearch-in-production/slides.html.

This was a very good event followed by a small chamber discussion. Thanks to all the folks who attended and asked very good questions :-) This was the first PJUG talk at the Egnyte office, I hope there will be more to follow.

Elasticsearch Caches - Fielddata

Mon, 06 Apr 2015 00:00:00 +0000

Exploring elasticsearch caches and memory usage - fielddata.

UPDATE this blogpost was written for elasticsearch 1.6 and it is no longer up-to-date.

One of my production clusters got recently hit by the CircuitBreakerException Data too large exception. In this post we’ll explore the Fielddata cache, Circuit Breakers and what we can do to avoid Fielddata eating all of our precious heap space. This a first article in a short series on elasticsearch memory usage.

Elasticsearch Workshop at Poznań University of Technology

Thu, 26 Mar 2015 00:00:00 +0000

I’ve conducted an Elasticsearch workshop at my university Career Fairs

I’ve conducted an intro workshop to Elasticsearch at Poznań University of Technology today. I was there because of Career Fairs where my employer wanted to attract young talents and make our brand more recognizable among students and graduates of technical majors.

The workshop group was very small - four students, but this allowed us to keep a good pace and we covered a lot of material. The interaction was also quite dynamic.

Make Me a Cluster

Mon, 23 Mar 2015 00:00:00 +0000

Easily provision an Elasticsearch cluster for fun, experimentation and profit.

If you followed my blog you may known than I’m responsible for changing my employer’s in-house search appliance to a new one backed by Elasticsearch. We had a huge data volume, even though we are not fully migrated to Elasticsearch yet. During our day to day operations we see a lot of interesting cases. I think we reached that point where reading documentation, news groups and blogs is simply not enough to explain everything we see in the trenches. I often feel a need to experiment with an Elasticsearch cluster.

Profiling Python Scripts

Fri, 16 Jan 2015 00:00:00 +0000

Run snake, run! How to profile python scripts with ease.

I have a python script which is responsible for a long running migration. The script communicates with three other systems - it reads some data from systems #1 and #2, merges them and then pushes them to the system #3. This is depicted below. The problem was that the migration run at a pace I wasn’t happy with. Since most of the work the script does it communication with external systems I wanted to know which is slow. Python has a great built-in profiler to answer this kind of questions. Follow this post to learn on how to use it.

Delaying Proxy in NodeJS

Thu, 08 Jan 2015 00:00:00 +0000

It is about time to add a new tool to your toolbox.

During the Christmas holiday season I had some time to learn NodeJS. In this post I’ll show how to create a simple (but useful!) proxy server in less than 30 lines of code. I’ll also give some hints on how to start with JavaScript with no prior experience.

Delaying Proxy

To give some background. I have a piece of code calling elasticsearch via its REST api and I wanted to test how it behaves in case of various network delays. I decide to create a simple proxy to simulate it. The idea is depicted below.

How do High Frequency Traders Make Their Money?

Sat, 06 Dec 2014 00:00:00 +0000

Michael Lewis offers a good story plus a layman introduction to high-frequency trading.

High Frequency Trading is quite a hot topic nowadays. It is a form of algorithmic trading where you gain an advantage by being faster than the other market participants. But how do high frequency traders make their money? The book Flash Boys by Michael Lewis, that I’ve recently read, is a good introduction to the topic. In this post I’ll summarize what I’ve learned from it.

Accessing Elasticsearch Cluster over https

Wed, 26 Nov 2014 00:00:00 +0000

Access your cluster securely in python or in java.

Elasticsearch offers no security out-of-the box. The connections are over http or their native protocol, both unencrypted, for all the world to see. This is not a problem if your cluster is in the same datacenter as the rest of your infrastructure - safely behind a firewall. Quite often this is not the case. You can put the cluster in Amazon EC2 or Google Compute Engine or something similar. If you do this, then need to encrypt the connection. Arguably, the easiest solution is to use https. There are various ways to configure the cluster, usually through some third-party proxy like nginx (please see the discussion). Recently, we’ve decided to spin a new cluster in Google Compute Engine and to allow only https access. I was curious if my existing java and python client code will work out-of-the box.

Searching for Product Name in Elasticsearch

Wed, 22 Oct 2014 00:00:00 +0000

How to implement good search on product name in Elasticsearch.

A lot of elasticsearch clusters will have a usecase of searching for product name. It doesn’t really matter if the products are consumer goods, articles or files. The important thing is that users what to search by product name and find matching items. The products should be found if a user types their exact name or just type something close enough. In the post I’ll describe a possible implementation of this usecase.

Elasticsearch Refresh

Mon, 20 Oct 2014 00:00:00 +0000

The problem with near-realtime search and update-by-query.

In the startup I work for we’re evaluating the usage of elasticsearch to search for file metadata. Recently, when writing an integration test I’ve found an interesting bug, which I missed when designing the solution. We use the update-by-query plugin to bulk update docs. In test I wrote I added a document and immediately updated it. The thing is that the document was not being update. What happened? Read on…

Two Recently Discovered Emacs Plugins

Tue, 30 Sep 2014 00:00:00 +0000

Edit your gmail messages with emacs and make http requests from your favorite editor

I recently found two interesting emacs plusing which can be applied for many usecases.

Gmail Message Mode

gmail-message-mode lets you invoke an emacs editor straight from the compose screen in gmail. You need to click on an icon or you can use a keyboard shortcut - alt + enter (this key chord is not customizable, its just on/off, what a misery for emacs geeks ;-)).

Standing Desk at Home

Thu, 31 Jul 2014 00:00:00 +0000

Use a laptop tray to create a standing desk at home with (almost) no effort.

Looks like this standing desk is not a niche anymore. There are lot of articles on-line about its benefits and issues the people are facing (like this or this).

Anyways, I decided the last year that I want to try it. A dedicated desk for standing can be quite expensive and also you may need to ditch your current desk if your home office is small. Of course, this was to be an experiment, so I wanted to try it with minimal effort and cost.

Host Gitlab on DigitalOcean

Tue, 08 Jul 2014 00:00:00 +0000

Building own github in the cloud.

I recently migrated my private repos to digital ocean and gitlab. This post is a walk through on how to do it and summarizes my experience.

Intro

Digital Ocean is a VPS provider. They have their data centers in NY, SF, Amsterdam and Singapore. Their servers are provisioned with SSD drives and have a very competitive pricing. They are a new kid in the block (launched in 2013), but they gained a lot of traction over the internet and there are a lot of success stories posted by different people. I think this is because they offer a good service, a simple UI and a competitive pricing.

CAP Theorem and Elasticsearch

Thu, 26 Jun 2014 00:00:00 +0000

Will it fail together with the network?

Let us check the Elasticsearch behavior in context of the CAP theorem.

CAP Theorem

CAP theorem stays that a distributed system communicating over an asynchronous network can’t provide these three properties at once:

Consistency: A read sees all previous writes;
Availability: Reads and writes succeed on all nodes;
Partition tolerance: These properties stay intact even in case of a network failure of some of the communication links between nodes.

This theorem is often used (and misused), but luckily there are some good clarifications over the Internet (see e.g. 1, 2 or even plain english, yet sexist and overly simplified explanation).

MapDB

Tue, 20 May 2014 00:00:00 +0000

Fast, off-heap java database with a collection interface.

The problem statement is this - you have a set of text files. Each row in the file represents a row from the database. The rows can belong to one of two tables (they can represent one of two distinct entities). There is a parent-child, one-to-many relationship between the entities. The parent always comes before its children, but you do not know how many children there are left. You need to denormalize the parent data into child entities and dump the data back to a text file. See the dumbed-down example below.

Samsung Fail

Thu, 24 Apr 2014 00:00:00 +0000

How not to update firmware + solution.

I own a Galaxy S3 smartphone, which I’m quite happy with. Android has its issues but the hardware is very good and samsung rolled a few apps to compensate for the OS shortcomings. However today I was really disappointed. They rolled out a new update. Updated 2014-05-06 with a solution.

Take a look at the change log:

Added new feature.

Really? Is this worth 50+mb of firmware update to add a feature? And then, take a look at the incentive for me to update:

Near real-time analytics

Tue, 15 Apr 2014 00:00:00 +0000

Big data = big fail?

Martin Fowler recently summarized a “Reporting Database” pattern. The core idea behind this pattern is to split analytical and transactional processing by providing a separate database for the former. A separate database for reporting purposes is nothing new and this pattern is widespread in the enterprise space. What caught my interest though was his opinion on near real-time analytics.

These days the desire seems to be for near-real time analytics. I’m skeptical of the value of this. Often when analyzing data trends you don’t need to react right away, and your thinking improves when you give it time for a proper mulling. Reacting too quickly leads to a form of information hysteresis, where you react badly to data that’s changing too rapidly to get a proper picture of what’s going on. Source

Java 8 Released

Sun, 30 Mar 2014 00:00:00 +0000

Whats new in Java 8? Where to start learning?

Last week we saw the new release of Java. Version 8 brings a lot of new features into the language. I think the most important are:

lambda expressions,
default method implementations in interfaces,
new datetime api.

There is also a lot of smaller improvements and additions in the standard library.

Image from http://www.jamesiliff.com/speechless-protagonists-spatial-storytelling-and-immersive-worlds-half-life-game-narrative-review-at-gdc-2012

I’m a huge fan of the functional programming paradigm and of course I’m very excited about the release. Hope it will be quickly and widely adopted in production. Also the library improvements will make a compelling alternative to popular external libraries like JodaTime and Guava. This is good, because I think everybody were including them, even for small projects.

Project Lombok

Thu, 20 Mar 2014 00:00:00 +0000

Never write your getters again!

My college recently introduced Project Lombok to our java code base. This is a great tool, which aims to make java source code less verbose by using a compile time annotation processor. In this post I give you a basic example of usage of Lombok and a description on how to configure IntelliJ Idea to use it.

Basic usage

To give a quick example (src)

This piece of code (notice @Data annotation) with Project Lombok

A Note on MongoDB for Java Developers Course

Fri, 28 Feb 2014 00:00:00 +0000

A study of a scandal with MongoDB.

The on-line course - MongoDB for Java Developers - I’d attended recently is finished. According to the course dashboard

Certification of completion will be available for download through your MongoDB Dashboard on or before March 10th.

This was a great experience, kudos for all the guys at 10gen who made it happened. I really liked the final exam, especially the first few questions. The dataset they operated on was the Enron Corpus. This is

Second Week of Mongo M101J Course

Sun, 19 Jan 2014 00:00:00 +0000

First impressions on MongoDB and M101J course.

I finished the second week of the MOOC MongoDB for Java Developers which I advertised in my latest post. I really like the course and I think its worth to spend some time learning Mongo.

MongoDB itself is an example of a NoSQL database and you can learn more during the course. In this post I want to highlight some features presented in first two weeks which I feel are inconsistent and surprised me a bit.

M101J - MongoDB for Java Developers

Mon, 06 Jan 2014 00:00:00 +0000

Free MongodDB course for Java Developers starts today.

10gen, the company behind MongoDB offers a free on-line course, which ends with a certificate. It’s a great opportunity if you want to get quickly up to speed with MongoDB. The course starts today and you can still sign-in.

The course is available here: https://education.mongodb.com/courses/10gen/M101J/2014_January/about

Learn everything you need to know to get started building a MongoDB-based app.

This course will go over basic installation, JSON, schema design, querying, insertion of data, indexing and working with language drivers. In the course, you will build a blogging platform, backed by MongoDB. Our code examples will be in Java.

Metaprogramming in Groovy

Sat, 07 Dec 2013 00:00:00 +0000

Groovy the language offers neat metaprogramming features. To metaprogram means to write programs manipulating other programs. In context of groovy this enables us to synthesize methods and classes, intercept calls and make objects behave like instances of different types. All of these manipulations happen at runtime. Groovy also offers compile-time AST (Abstract Syntax Tree) manipulation.

Before we move out to the features themselves, let’s start with figuring out how Groovy can change behavior of various classes of objects.

Personal Kanban

Thu, 29 Aug 2013 00:00:00 +0000

Productivity system inspired by lean and agile ideas which helps you focus on what is important.

Recently I’ve read an interesting book by Jim Benson & Tonianne DeMaria Barry - Personal Kanban. This book not only promises to be 100% new age free, but also not to focus on doing more. Sounds unorthodox for a productivity book? Well, it surely does sound interesting. This post is intended to be a short description of the Personal Kanban (PK) process - it should contain enough information to get you started. Please bear in mind that it is also an enthusiastic description - I decided to summarize PK and not review it, yet. However, in a few weeks I plan to write a companion post to describe my experience with implementing PK at home and in the office and evaluate it it more details.

Markdown in the (Browser) Box

Tue, 13 Aug 2013 00:00:00 +0000

Clean and good looking emails including code & math. Write them using markdown and render to html within your browser.

You need to write a long email. With all the sections, sub-section, lists, etc. You need to put some code there (or just let anybody know what God said to make the light). Luckily for you, html emails were invented - you can open the gmail UI and start typing… Wait. Did I mean gmail html editor? The one that needs three clicks to monospace a piece of code? There has to be a better way. And there is - use Markdown Here. A plugin to your browser which lets you write emails in Markdown and then render it with a key stroke.

Remote Debug Junit Tests Run From Ant

Thu, 08 Aug 2013 00:00:00 +0000

How to turn on remote debugging for junit testrunner in ant.

Sometimes you want to run a testset in an isolated environment. Be it a different box, a virtual machine, a different user or just a console and not an IDE. Reasons may vary, e.g. you need a special setup for the integration tests or you have a fancy build configuration which is hard to reproduce in the IDE. Still, no matter what your reasons are, you want to debug such tests. In this post I’ll show how to use a remote debugger to connect to a test environment using IntelliJ Idea or Eclipse. I assume you have an ant build.

Ubuntu on Samsung Series 7 Ultrabook 730U

Mon, 22 Jul 2013 00:00:00 +0000

Is it possible to install Xubuntu 12.04 on Samsung NP730U3E?

I’ve got a new laptop - Samsung 7 Series NP 730U 3E. It’s 13.3’’ ultrabook with a quite decent hardware under the hood:

Intel Core i7-3537U,
10 GB of RAM,
13,3’’ Full HD Display (1920x1080),
128 GB SSD drive,
Two graphic cards: AMD Radeon HD 8570M + Intel HD Graphics 4000.

This machine comes with a preinstalled Windows 8. It is a bit of an issue for me - I heavily rely on linux both for work and for my other projects.

Be Your Own Certificate Authority

Sun, 21 Apr 2013 00:00:00 +0000

How to create a Certificate Authority and use it to generate SSL certificates.

There are many reasons why you may need to create your own CA. Although you are advised to sign internet facing services with a certificate from a trusted root entity you have much more freedom when it comes to your intranet or vpn infrastructure. The other use cases may include a WPA2 enterprise access point or an https connection to a private wiki page.

Legal

Mon, 01 Jan 0001 00:00:00 +0000

Licence #

All the content of https://kupczynski.info is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License unless stated otherwise.

Disclaimer #

The views and opinions with stated on this blog represent my own and not those of people, institutions or organizations I am affiliated with unless stated explicitly. Especially my blog is not affiliated with, neither does it represent the views, position or attitudes of my employer(s), their clients, or any of their affiliated companies.

Home on Igor Kupczyński

YOLO Mode in a Vagrant Box

The problem #

Solution: disposable VMs #

Pattern Spotting: Hatchet

PostgreSQL as the Universal Backend #

GCP Live-Migrates, AWS Reboots: How Cloud Providers Handle Host Maintenance

The incident #

TCP connection limiting in Go: Blocking Accept vs Active Load Shedding

TL;DR #

AGENTS.md: A Standard for AI Coding Agents

The Problem #

Why AI Agents Hit the Wall

Compounding Works Both Ways #

Effective Prompt Engineering in 2025

Why Prompt Engineering Matters #

Core Concepts Covered #

Raycast browser switcher: a vibe coding rabbit hole

Vibe coding to the rescue #

Principal Engineer Roles Framework

The Six Principal Engineer Roles #

1. Sponsor #

DNS failover: When zero-weight records strike back

TL;DR #

Chain of Density: An LLM summarization technique

Founder Mode

Paul Graham’s “Founder Mode”

Amazon S3 now supports conditional writes

TIL S3 supports conditional writes

Amp It Up

Make Go application container resource limits aware

TL;DR

Memory

Golang 1.20 upgrade failed due to GLIBC_2.34 not found

Negotiate your salary

Private Link is the IP filtering of the cloud

Private Link #

The Phoenix Project

Fun with docker networks

The story #

Notes on creating a telegram bot

The problem #

Remote debug your go code

Remote debug #

Natural project planning

FIPS compliant crypto in golang

Zero to One

Background #

Elastic Management Philosophy

Server Name Indication with Traefik and Go

Server Name Indication (SNI) #

The Slowest Hiker

Set up AWS Lambda with terraform

AWS Lambda to invalidate CloudFront when S3 object changes

Problem statement #

OpenVPN and DNSes on Ubuntu

Fire and Motion

Change the maximum number of open files in ubuntu 18.10

Nvidia on Ubuntu 18.10 requires lightdm

Problem #

Use wine to run windows JDK on linux

Diagnose SSD failures

Ubuntu 18.04 on Thinkpad P51

Zoom and Kubuntu 18.04

JMH - Java Microbenchmark Harness

Getting started #

Who's Got the Monkey?

JVM Class Data Sharing

What is Class Data Sharing? #

Notes from Geecon 2018

Learning VIM with Spacemacs

Evil-mode #

12 Rules For Life

Background #

IT Career Options

Shell in Emacs - Performance

How fast can you print? #

Insanely Simple: The Obsession That Drives Apple's Success

Background #

Elasticsearch and absolute_form in HTTP/1.1 request line