toad.social is one of the many independent Mastodon servers you can use to participate in the fediverse.
Mastodon server operated by David Troy, a tech pioneer and investigative journalist addressing threats to democracy. Thoughtful participation and discussion welcome.

Administered by:

Server stats:

269
active users

#containerization

1 post1 participant0 posts today

Understanding Podman Coming from Docker

This is going to be a bit of a ranty post, but engineers are good at that kind of thing. So here goes.

A bit of Background

I started my devops career learning about docker and docker-compose. I enjoyed the elegance of being able to write a file of what services I want in a yaml file and bring the entire thing up. Very very easy to rapidly deploy something for testing and into my own production. Use this image, on this port, here’s where to store your data and you can connect to this database. A simple compose file could look like this.

services:  memos:    container_name: memos    image: neosmemo/memos:stable    ports:      - 5230:5230    volumes:      - ./memos/:/var/opt/memos    restart: unless-stopped

Download and run this webapp called memos, use port 5230, store your data in a directory called memos. Restart if anything crashes until I tell the daemon to stop it. Very easy to understand and follow along!

Docker has a few cons

Docker has a few pain points however. The biggest one being that it runs a centralized daemon and if you need to update docker, e.g. a security update is released well… all of your services have to come down for a moment while the patch is applied. That kind of sucks in terms of maintaining stable services and if you have your server set to automatically apply updates (remember automation is good!) who knows when your stuff can go down.

Docker also doesn’t come in the default Debian/Ubuntu repositories. You have to install it from a third party repo. This isn’t that big of a deal but the additional configuration is just something you have to remember when setting up a new test environment instead of a simple apt install docker.

Lastly docker does support a rootless mode, though it has some limitations. This leads me to my search for alternatives, wondering if there might be something better out there.

Enter Podman

So I decided to examine podman. I wanted to find a way to follow my same workflow (or as close as possible) while using this different implementation of containerization. It addresses my concerns: it comes default in the repos, it natively supports rootless without any monkeying around, and it is daemonless! Sounds good right? Not exactly.

Podman-Compose

Okay so as it turns out someone thought of this already! A python program to implement the same thing as docker compose except through podman as your backend. Perfect! Except… every time I try to run a stack I get python errors and crashes…. unless I run it as sudo! So much for trying to escape the need for root there. I’ve tried testing it with a simple deployment of bookstack with a backend mariadb. Never works right! I dont think it properly creates the environment variables defined in the yaml file. So really what’s the point of fighting this? Next!

Podman generate/play

So I start searching around to see if there’s a more native implementation of compose like files with podman and I find this redhat article! Written by the people who made the damn thing I expected a nice outline of how to do this. Holy shit was I mistaken reading this:

https://www.redhat.com/en/blog/compose-podman-pods

This is about as clear as mud. Basically the idea of podman-compose is still being worked about in their development. (To be fair this might have changed in the 5 years since this article was written) but they continue with the concept of podman “pods” and the command “podman generate” which will create a kubernetes style yaml file of what you’ve already created.

Well that’s nice but it seems completely ass backwards from what I’m needing here.
It seems I’m not the only one who’s had a negative experience:

Comment
byu/avamk from discussion
inpodman

I could continue trying to make this work, but at this point it seems like I’m fighting a fairly pointless battle. Yeah its daemonless, but my workflow isn’t compatible exactly. I could try systemd integration but that’s something I already barely tolerate as it is. I don’t see the point of continuing to fight this when docker already is perfectly functional as it is with a few minor nags.

Maybe someday Podman will be able to compete better on this front but for right now, it’s not for me!

www.redhat.comMoving from docker-compose to Podman podsIt feels like forever since I wrote my Red Hat Enterprise Linux 8 Beta intro to Podman. In fact, it's been quite a while, and a lot has happened sin...

Dear #Fediverse, I need your recommendation:

I'm building a #website that in the end consists of a #Docker #NodeJS image/container and a #Postgres container. What options are there to host those containers on the #Web? I'm looking for a cheap or even free (?!) solution. I expect less than 50 users, so big hardware or scalability is not an issue. Would a virtual server with #minikube be a viable option? I'm a Docker and #Kubernetes newb so please bear with me.

#Docker alike container management on #FreeBSD? Let's have a look at #Kleene - which works very similar to Docker and makes the shift for Linux users even easier to #Jails without having deeper knowledge but by benefiting of all the features on Jails.

#klee #kleened #bsd #runbsd #linux #container #howto #devops #community #jails #freebsdjail #container #containerization #jailmanager #manager #jls

gyptazy.com/howto-kleene-as-a-

This is a customer-facing role, so if that's not your thing, keep scrolling.

TLDR: If you know Hadoop and live close enough to Belfast to commute, you should apply.

I've posted this before, but it's been a little while #fedihire. Also, adding some additional information this time. This is my team. We are already on three continents and 6 timezones, but #Belfast is a new location for the team. I know literally nothing about the office.

I know a lot of places Hadoop is the past, and sure we see a ton of #Spark (I do not understand why that is not listed in the job description but maybe because they want to emphasis that we need hadoop expertise?). You can see all the projects we support at openlogic.com/supported-techno

It depends on how you count, as I was on two teams during tradition, but I've been on this team for over 5 years now. It's a great team. I've been with the company now right at 7 years. I cannot say how we compare to Belfast employers but this is well more than double where I have stayed at any other employer (even if you count UNC-CH as a single employer rather than the different departments, I've beat them by well over a year at this point).

My manager has been on this team for almost 15 years. His manager has been with this team for almost as long as me, but with the company much longer. His manager has been here almost as long as me (I actually did orientation with him). His manager is a her and she's been here almost as long as me. So, obviously, this is a place where people want to stay!

Our team has a lot of testosterone, but when I started, our CEO was a woman. The GM for the division is a woman.

My manager is black. The manager of our sister team is black.

I think you'll find our team and company is concerned about your work product and not how you dress, what bathroom you use, or the color of your skin.

If you take a look at our careers page, you'll see this:

Work Should Be Fun
There’s always something to look forward to as a Perforce employee: scavenger hunts, community lunches, summer events, virtual games, and year-end celebrations just to name a few.

We take that shit seriously. Nauseatingly so sometimes, lol.

Actually, we take everything on the careers page seriously, but I know from experience that some places treat support like they are a shoe sole to be worn down. Not so here. It's not all rainbows and sunshine, of course. The whole point is that the customer is having an issue! Our customers treat us with respect because management demands that they do.

------

The Director of Product Development at Perforce is searching for a Enterprise Architect (#BigData Solutions) to join the team. We are looking for an individual who loves data solutions, views technology as a lifestyle, and has a passion for open source software. In this position, you’ll get hands on experience building, configuring, deploying, and troubleshooting our big data solutions, and you’ll contribute to our most strategic product offerings.

At OpenLogic we do #opensource right, and our people make it happen. We provide the technical expertise required for maintaining healthy implementations of hundreds of integrated open source software packages. If your skills meet any of the specs below, now is the time to apply to be a part of our passionate team.
Responsibilities:

Troubleshoot and conduct root cause analysis on enterprise scale big data systems operated by third-party clients. Assisting them in resolving complex issues in mission critical environments.
Install, configure, validate, and monitor a bundle of open source packages that deliver a cohesive world class big data solution.
Evaluate existing Big Data systems operated by third-party clients and identify areas for improvement.
Administer automation for provisioning and updating our big data distribution.

Requirements:

Demonstrable proficiency in #Linux command-line essentials
Strong #SQL and #NoSQL background required
Demonstrable experience designing or testing disaster recovery plans, including backup and recovery
Must have a firm understanding of the #Hadoop ecosystem, including the various open source packages that contribute to a broader solution, as well as an appreciation for the turmoil and turf wars among vendors in the space
Must understand the unique use cases and requirements for platform specific deployments, including on-premises vs cloud vs hybrid, as well as bare metal vs virtualization
Demonstrable experience in one or more cloud-based technologies (AWS or Azure preferred)
Experience with #virtualization and #containerization at scale
Experience creating architectural blueprints and best practices for Hadoop implementations
Some programming experience required
#Database administration experience very desirable
Experience working in enterprise/carrier production environments
Understanding of #DevOps and automation concepts
#Ansible playbook development very desirable
Experience with #Git-based version control
Be flexible and willing to support occasional after-hours and weekend work
Experience working with a geographically dispersed virtual team

jobs.lever.co/perforce/479dfdd

www.openlogic.comSupported Open Source Technologies | OpenLogicOpenLogic by Perforce supports hundreds of open source technologies. Search this list to see which packages we most often support for customers.

Open Source replicability can benefit greatly from package managers like #Guix or #containerization. However, these technical tools should not obscure the fundamentals: your code must be thoroughly documented to be replicable. A README + codemeta + Citation file are highly recommended. This has been true since 2016, and nothing has changed in that regard since then. arxiv.org/abs/1607.01191 "Documentation is the glue that binds a data science project together" (ZIemann & Poulain, 2023)

I just switched my home server to #Gentoo after roughly 4 years of running #OpenBSD.

I loved running OpenBSD and have no regrets. It definitely forced me to develop good #sysadmin habits that I wouldn’t have learned easily elsewhere.

Making the switch definitely gives me an appreciation for the BSD model. #Linux truely feels like a complete mess, and I feel that way despite using linux as a desktop OS for roughly 20 years. It’s no wonder that Linux server management has evolved towards #containerization.

Here's a tutorial I wrote a few months ago that's aimed at software developers who work professionally with programming languages like Python, but aren't exactly familiar with Kubernetes and containerization. Essentially, it explains how to containerize a simple Python application in Docker and then deploy it to a Kubernetes cluster.

I have a few other technical writing pieces that I plan to make publicly available on my GitHub repository. In the medium term, I plan to create my own personal website just to store my portfolio of technical writing articles and to share/comment on interesting things I happen to notice about this profession. At the moment I'm still hesitating about the right technology stack to use - Hugo/Gatsby/Docusaurus/GitBook/Jeckyl...

"This hands-on tutorial provides a step-by-step guide to preparing an application image, setting up the necessary infrastructure, and deploying the image to a Kubernetes cluster. It is written for professional software developers who are new to Kubernetes and containerization."

github.com/remixtures/nordlys/

GitHubnordlys/Deploy_Your_First_Application_Image_To_Kubernetes/README.md at main · remixtures/nordlysTutorial on how to deploy an application image to Kubernetes - remixtures/nordlys