Talk with your Kubernetes logs with natural language ( AI-driven K8S operator )

Can you talk to your Kubernetes cluster using natural language? Yes! I've implemented the simplest AI-powered interaction with Kubernetes to inspire others to explore this path further—or even transform K8S Whisperer into the Tony Stark of Kubernetes management. 🚀


Demo video :

https://www.youtube.com/watch?v=T3E9Wjbq44E&list=RDsa7uGYm-ixA&index=25


Source code :

https://github.com/ARAldhafeeri/K8sWhisperer-




https://redd.it/1im6tmo
@r_devops

I created 3 FREE AWS Practice Exams w/ hundreds of random questions to help you ace your certification (SAA, Cloud & AI Practitioner)🎯

I'm excited to share a comprehensive AWS certification practice pack with you! As someone who has navigated the AWS certification journey, I understand the importance of having access to quality study materials. That's why I've created this free resource pack featuring three complete practice exams:

You can access all three practice exams here

AWS Cloud Practitioner
AWS Solutions Architect Associate
AWS AI Practitioner

Each practice exam features hundreds of carefully selected questions covering all essential exam topics and domains. You can choose between two formats:

Basic mode: 35 questions, 40-minute duration
Full mode: 65 questions, 90-minute duration

Key Features:

Real-time score tracking during the exam
Detailed answer review to learn from your mistakes
Randomized questions for more effective studying
Comprehensive coverage of all exam domains
Matches the real exam format and difficulty

While these practice exams are valuable study tools, remember that hands-on experience is crucial! I highly recommend complementing your studies with AWS Skillbuilder for practical experience.

I developed these practice exams with dedication and care to support our community. While you'll find information about contributing to the project within the links, rest assured they will always remain completely free, regardless of contributions. I believe quality AWS certification preparation should be accessible to everyone!

Want to stay updated on future resources? Connect with me on LinkedIn!

https://redd.it/1im9y66
@r_devops

K8s CD tools where spoke clusters create connection to hub cluster

I'm investigating open source CD tools to deploy apps on multiple clusters running on IoT devices. We're considering something similar to a traditional hub-and-spoke pattern, but where the K8s agent/operator on the device cluster initiates the connection to the hub CD management plane. That means the hub no longer needs ingress to the devices hosting the cluster.

Does anyone know of CD tools that work this way? I have found ArgoCD Agent (https://github.com/argoproj-labs/argocd-agent), but that is still experimental. We're not married to GitOps tools, so open to alternatives.

https://redd.it/1im9j5m
@r_devops

Externalizing pipeline and making it consumable

Good news / bad news

Current application owners love my new pipeline….automated huge portions of the build and deployment process, I even built custom pieces to create RFCs 💅🏻

Bad news, entire org wants to move to my pipeline

So… for those who have done something like this. How do I do this without losing my mind?

I want to move individual steps out, token rotations, security scans, build steps… etc. Move them one at a time and make them consumable…?

Current application using gradle… some use
Maven… some both lmao

Was literally just told “You choose how to handle it”…

So… help? 😅😅😅

https://redd.it/1imazty
@r_devops

Are there versioning tools for deploying multi-service mono-repos?

I've been confronted with a reoccurring challenge where I have a multi service software architecture where all the services are sourced from the same git repository. These services are all deployed as Docker images to a Kubernetes cluster.

The challenge I have is that I only want to rebuild and redeploy the particular services that have actually changed. I also need to handle cases where one service may reference a common library which can also change.

So, let's say I have three services A, B and C. A and C both reference common library lib-FOO. So, for example:

If C's source code changes, I want C to be rebuilt and deployed whereas nothing happens to the other services ( A and B).

If the source code to lib-FOO changes, lib-FOO, A and C all need to be rebuilt and redeployed. B hasn't changed, so no actions should be taken for it.

Are there any specific tools or technologies I can use for this scenario?



https://redd.it/1ime7kj
@r_devops

What should a transitioning engineer know to be successful in dev ops?

Greetings,
I am a systems engineer working in defense, who also has experience in the embedded world. I am considering moving industries. (Never mind why)

I was taking with a friend of a friend the other day who works in back end web development. Going over my skills I was rather surprised that they suggested dev ops as a possible new role. Their reasoning: I often take old software and create integrations to keep them running in modern environments. Sometimes I use VMs, sometimes using docker/podman, and sometimes occasionally just recompiling the code with small changes to the code/build scripts. (This isn't my main role, just something I get tasked with regularly.)

Long story short, what kind of skills would look good to an employer for someone transitioning into this field? I.e. with experience but not directly related experience. Any certs or online classes worth checking out, whether for the resume or for practical knowledge?

https://redd.it/1imbt93
@r_devops

How often do you consult other team members?

TO preface, I am junior and I was lucky enough to land a role in this. But, i have been “let loose” so to speak, barely any oversight but at the same time. I dont have anyone to talk to or bounce ideas off of, other than my boss - who is the lead.

Got a recent request from the Dev team to implement some stages in our pipelines and it will require a good amount of net new changes. I have an idea as to what I need to do, dont have it all yet figured out but will go step by step(Yes I know 90% of you are experts, i am a novice.)Wanted to pull in the boss man, or just ping him and let them know of my recent request, and maybe jot down some high level ideas of next steps. Boss is a busy person, but usually answers my questions, albeit , very brief responses.


Would this be a bad look to even reach out and say “ Hey boss man, got this recent request which came in, have an idea of how to go about it, but wanted to bounce some ideas off of you” or even just provide a high level outline of what i would be working on . Trying to move up to a mid-level and dont want this to mess my chances up. I’m probably freaking out over nothing but still…

https://redd.it/1imh1af
@r_devops

Seeking Advice on Managing Environment-Specific Configuration without changing code

I’m working on a Next.js project deployed to OpenShift using Docker, with five distinct environments. The issue is that the project needs pre-build configuration, which includes environment-specific variables that are injected in the entry point with a prebuild Node.js script. Every time there’s a change in the environment (like modifying an ingress in a deployment), I need to adjust the configuration file for that environment and redeploy.

I can’t use OpenShift’s configMaps or secrets because they’re meant for runtime, and I don’t want to have a separate Dockerfile for each environment. So, I’ve considered two possible solutions:

1. Create an API that provides the configuration values before the build, basically a Service Discovery pattern. This way, in the script that runs before the build, I can make a request to get those variables. I’ve heard HashiCorp Consul could be used for this, but I’m looking for something free.
2. Do the build with placeholders for those values (like the environment's hostname) and use a script with envsubst in the Dockerfile before starting the container to replace the placeholders. This would involve analyzing many files during the build.

Both solutions would require a redeployment, but at least I wouldn’t need to modify the code directly. Plus, the API solution could work for other projects that also need environment-specific variables during the build. The purpose of this post is to get some feedback, as I’ve been thinking about this problem for so long that I might be overcomplicating the solution.

https://redd.it/1imhsq1
@r_devops

Kubernetes is the new Jenkins....

With all the operators, extensions and nightmare of keeping it up to date kubernetes is the new Jenkins....

https://redd.it/1imk1ry
@r_devops

NOC to DevOps or Cybersecurity?

Hey guys, a little bit about myself, I'm turning 22 soon, I've started my career through Tier 2 Support Specialist for 2 Years, made a transition to NOC within the same company (Present) currently I'm in this position for 4months.

This department will be closed In May and I don't know which path to choose.

NOC > Junior DevOps or other similar rules, (Existing experience In DevOps tools and cloud while I keep learning during my work)
NOC > SOC / Incident Respone Analyst (where I need to learn from scratch but I've always had passion for ti)

If you guys were in my shoes, which path you'd choose or what you'd do if u were me?

If there are any more similar rules feel free to list them here.

Thank you guys it means a lot!

https://redd.it/1imkv5l
@r_devops

Anyone cares for Datadog Vector as SaaS?

I see developers and DevOps struggling with running in-house OTel pipelines.
Would you like to subscribe to a SaaS version of Vector (https://vector.dev/)?

The only prohibitive cost would be the outbound cost, which can be offset if the SaaS service provides a CloudFront link to send the data to since the cost of EC2 to CloudFront is USD Zero.

Would you still not use this service, and why?

https://redd.it/1imjuas
@r_devops

How to Publish to GitHub Pages From Another Repository

Hey DevOps folks!

I wrote a detailed guide on deploying static sites from one GitHub repository to another using GitHub Actions and OpenTofu.

This setup is particularly useful if you want to:

- Keep your source code private while using free GitHub Pages hosting
- Manage infrastructure as code using OpenTofu/Terraform
- Automate cross-repository deployments with GitHub Actions

The guide walks through:

1. Setting up the target GitHub Pages repository
2. Configuring the source code repository
3. Creating necessary deploy keys and GitHub Actions workflows
4. Implementing the deployment pipeline using OpenTofu
5. Managing the infrastructure with Terragrunt

All code examples are provided, including complete GitHub Actions workflows and OpenTofu configurations.

https://developer-friendly.blog/blog/2025/02/10/how-to-publish-to-github-pages-from-another-repository/

Let me know if you have any questions!

Please share in the comments if you prefer an alternative approach.

https://redd.it/1imb0fy
@r_devops

YAHH - Per-project history file

YAHH is a Zsh-based tool that helps you manage separate command histories on a per-project basis. Instead of having one global history file or one per directory, YAHH allows you to keep distinct histories—called realms—for each of your projects.

This makes it easier to recall recurrent commands that are specific to a given project or operational environment, useful in professional services, consulting and other context-switching role.

https://github.com/Positronico/yahh

https://redd.it/1imt4fz
@r_devops

WebUI for Ansible with some monitoring?

A while ago, I used Puppet and Foreman, and it was quite practical. I used it for the whole pipeline, from setting up hosts to configuration management. However, I now use Ansible more and more, and I find that Foreman doesn’t integrate well with Ansible. Also, Ansible Tower/AWX is not very comfortable to use. So, I mainly rely on the command line, which works well enough. That said, I was wondering if there is a modern GUI for Ansible that also includes basic monitoring features, similar to Foreman, where Facts where visualized.

https://redd.it/1imvqiw
@r_devops

Implementing deployment windows

Hi,

I've been asked to look into managing some code releases (across various systems and methods) through specific pre-arranged release windows and I'm trying to find information about good ways to do this, pros and cons etc. However searching online is a problem because I just get endless results about deploying MS Windows changes 😂

Can anyone point me in the direction of some good resources or feel free to reply with opinions? I'm obviously aware this is going to work better for some types of change than others.

https://redd.it/1imwkat
@r_devops

List of 650+ well-funded startups that don't suck (remote, US, EU)

Finding well-funded, growing startups with strong engineering/product cultures is really hard. Created www.startups.gallery to make finding them easier. And no, this is not another spreadsheet or pay-to-play directory. It's just a thoughtful collection of today's most interesting projects, curated by humans. And yes, I know that startups aren't for everyone, but these are hopefully the most promising ones. Open to all and any feedback!



https://redd.it/1imzow5
@r_devops

Confused and scared please help!!

Hi guys
I am an AWS Architect with pretty good experience. I recently switched companies. I had interviewed for an Architect role. I got assigned to a project and it's AWS Devops. I got the initial KT and feel like a fish out of water. I am scared if I can cope up and do well. Please help me I am literally going thru mental breakdown....
I don't know if I should ask for another project or what?!

https://redd.it/1in1w4n
@r_devops

How would you architect an an analytics webapp?

I architecting an app that will have terabytes of ecommerce product data (products, titles, review) that is being scraped every 15 minutes and stored into S3 as JSON. The clients in this case will be a web app (mobile app in the future) that shows trends over time (such as price trends), number of reviews, etc. Users will be able to select date ranges and other ways to filter and aggregate the data. Frontend is written in React, backend and pipelines are written in Python.

I am torn between the following:

1. just simply doing ETL from S3 into TimescaleDB, which the apps will query.
2. doing ETL from S3 into Clickhouse for analytics, and a separate Postgres instance for relational queries (users accounts)
3. Implementing a lakehouse architecture by using Apache Iceberg on top of S3 and seeing if the app can use a Python Trino client to query. Lakehouse architectures seem more efficient than traditional data warehouse architectures - I'm just unsure of the performance within applications.

How would you build this?

https://redd.it/1in1n93
@r_devops

I ask ChatGPT about AI replacing DevOps engineers every quarter and this time it finally gave me an answer

Question:

>How long does it take till AI replace the majority of devops and sre engineers. Give me a number not the stuff like AI won't replace bla bla bla

Answer:

A realistic estimate is **5 to 10 years** for AI to replace the **majority (over 50%)** of DevOps and SRE engineers.

* **1-3 years**: AI will significantly **augment** DevOps/SRE roles but not fully replace them. Expect automation of routine tasks (CI/CD, monitoring, incident response).
* **3-5 years**: AI will handle **most** of the operational workload, reducing team sizes. Companies will rely on AI-driven platforms for reliability engineering.
* **5-10 years**: AI will likely **replace** more than **50% of DevOps/SRE engineers**, keeping only a small number of experts for oversight and edge cases.



https://redd.it/1in6ual
@r_devops

How Do You Track Developer Environment Costs?

Hiiii, so I'm doing some research on how companies approach developer budget costs, specifically around environments. Sorry, feel free to brain dump, just working to understand how others might do it differently than me.

For those of you managing a dev team and balancing environment costs—do you track expenses per developer or per environment? Do you assess costs on a monthly or yearly basis? And do you try to maintain a certain ratio between staging and production costs?

For example, at my company (on the small side) we track staging vs. production costs monthly but don’t break it down per developer. I’m curious if larger companies handle this differently—or if other small companies do things another way.

Hit me with your thoughts :)



https://redd.it/1in7svm
@r_devops

Does anyone have a AWS CDK example of a ALB + ECS Fargate + CodePipeline blue/green setup?

I am really struggling to find a holistic example of this in documentation or elsewhere. I'm CONSTANTLY running into a chicken or the egg scenario between ECS and CodePipeline. In click-ops I can get it working almost instantly but its proving to be a serious pain for me in my AWS CDK IaC project. Feel like I've tried a million combos but nothing has worked E2E yet.

Note: I'm talking about a full ECS Fargate + CodePipeline (+ source, build, deploy) setup btw - where we have the task defs/appspec in the source repository, then want to fetch and use them as well as ECR image during each pipeline execution.


I know this is from a knowledge gap on my end, any help is super appreciated.

https://redd.it/1in90we
@r_devops