Load shedding choice

Hey all,

So we've got a pretty usual stack, AWS, EKS, ALB, argocd, aws-alb-controller, pretty standard Java HTTP API service, etc etc.

We want to implement load shedding with the only real requirement to drop a percentage of requests once the service becomes unresponsive due to overload.

So far I'm torn between two options:

1) using metrics (prom or cloudwatch) to trigger a lambda and blackhole a percentage of requests to a different target group - AWS-specific, doesn't seem good for our gitops setup, but it's recommended by AWS I guess.

2) attaching an envoy sidecar to every service pod and using admission control filter or some other filter or a combination. Seems like a more k8s-native option to me, but shifts more responsibility to our infra (what of envoy becomes unresponsive itself? etc).

I'm leaning towards to second option, but I'm worried I might be missing some key concerns.

Looking forward to your opinions, cheers.

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

My current title is “Senior Software Engineer” but I do DevOps. I am getting zero bites on my resume. Is it because of my title?

I work at a software company in the Platform Deployment organization. We do EKS, Terraform, Jenkins, Grafana, etc… My daily job is DevOps. However, everyone in Platform Deployment gets the title “Software Engineer”.

I am a senior level, 15 YoE since college, trying to find a new job, but getting zero bites on my resume. Is it because my current title is not DevOps-related? I know the market is bad, but surely not this bad, right?

https://redd.it/1myy29h
@r_devops

Which AWS service for streaming voice + text to AI providers?

Greetings fellas,

I want send a voice recording along with some text to an AI provider. Will stream from the user's computer & also with an HTTP request backup.

>User computer >---stream/http--> AWS >---http--> AI provider
‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ ‎ |
User computer <--------http-----< AWS <--------http----/

My Question is, Which AWS service is best suited for this?

AWS will be there as the middleman to authenticate the request, process it and then return the response. Problem is I saw that there is a payload limit of 6mb with Lambda functions. The first stream/http will easily be over 6mb manytimes :( So would need something that accommodate more requests at least 10 - 20mb.

User authentication is already implemented using Supabase. I can't use supabase edge functions for the above though because of the delay. I got the 200$ AWS free trial haha 😂

Your kind advice is highly appreciated <3

https://redd.it/1myz7ci
@r_devops

4 yoe in production support, observability and disaster recovery management, should I give up the ambition to move to DevOps?

I'm stuck in tutorial hell and idk where to go from here because I'm burnt out because of work and upskilling - should I just give up and move to something less dense? I chose DevOps because I've been great at solving prod issues and getting it working asap and plus I understand the ideology but the plethora of tools is making it tough, I'm through the basics, linux, networking, shell and what not, I know the roadmap I know what I need to do but idk if I have the strength to do it, what do you guys think?

https://redd.it/1mz12gj
@r_devops

Career switch

Hello everyone! Need an advice from experienced people. Few weeks ago I was working on a project as QA engineer (Python). Before that I almost have never talked to DevOps engineers and didn’t even think about what they do, but then I’ve discovered a lot of interesting stuff about DevOps. I really felt that I would really like to dive into it and probably do career switch. But I don’t really know what would be the best way to start my way. I know Python enough to write auto tests for API and with Playwright, and I heard it’s one of best languages for DevOps. In general, I have small experience with Docker and Linux (Month ago I installed Ubuntu as a main OS, but now I think about arch), know basics of networks and git.

But one of biggest problems - I don’t have an experience of development, I worked only as a QA, and I only studied development, not worked as a developer.

Anyway, I don’t really know what would be the best way to start in my situation. And is this even possible, what do you think?

https://redd.it/1myz69b
@r_devops

"What are the essential and even the underrated VS Code extensions that are widely used by devops/cloud?"

How many extensions do you use for devops usecases?
which extension you think is the best that everyone should use in day to day life??

comment your favourite extensions that you cant't live without it

https://redd.it/1mz3ihy
@r_devops

Building an AI-Powered Compliance Monitoring System on Google Cloud (SOC 2 & HIPAA)

GCP compliance monitoring system by implementing a multi-agent setup using the crewai\_coding\_crew template from the Agent Starter Pack.

https://medium.com/@rasvihostings/building-an-ai-powered-compliance-monitoring-system-on-google-cloud-soc-2-hipaa-eecf7a5c30e4

https://redd.it/1mz8w0r
@r_devops

Best way to approach DevOps?

Pardon me if this question has already been asked.

I am looking to venture into the world of DevOps. I am quite a dedicated learner and I have a deep passion for the field and as such I should have no problem learning the same and putting it to practice. However, it is to my understanding that DevOps is not a Junior Role and as such I have decided to start learning System Administration then make my way up from there. My question is, is it feasible to focus entirely on System Administration courses for starters then learn the rest of DevOps content while carrying out sysadmin practices or I can just learn DevOps and still be knowledgeable enough to land a junior sysadmin role?

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

what's the point of devcontainers if it forces standardization?

I was under the impression that devcontaineres are a way to:

-> Specify a base image for everybody
-> If you use vscode, great, semaless layering. Else, add a few post-install scripts to drop waht you need into the project, great.

Then you have a base container with just only the needed build dependencies.

But like in practice, there's no such thing as "version controlled developer specific devcontainers", you have to centralize while not checking in core changes to the main repo.

My neovim setup might tick off, either somebody else's vim setup or someone who just wants vscode and nothing else (and I certainly don't want to pull in extra vscode stuff). Our project might not use python, but for my tooling I will get python anyways (neovim stuff). perfect use for devcontainers, I just install python post-install but committing that to the main repo? That... can easily get cluttered, if other people's devcontainer configs are there too.

Also, there are some user namespacing things (set uid gid etc) that might not belong in a build container but we'll worry about that later (also I think in practice you should have a separate deploy container, but we have not gotten to that level of automation yet, we have just started using build containers lol).

I guess, from more experienced developers, how do you deal with this issue where 5 different people have 5 different ide/dev workflows and you want minimal frictoin integrating with a devcontainer workflow?

https://redd.it/1mzc191
@r_devops

Starting My Cloud Career — Which Certification Should I Pursue Next? Terraform, Kubernetes, or Azure Cloud Architect?

Hi everyone,

I'm at the beginning of my cloud career and looking for advice on which certification I should pursue next to boost my skills and job prospects. I have around 13 years of experience in telecom, networking, and IT fields, and a little hands-on cloud experience. I also have a solid foundation from my current certifications, which include:

* AWS Certified Solutions Architect – Associate
* Multiple Azure certifications (3x)
* Alibaba Cloud certifications (3x)
* Oracle Cloud Infrastructure Generative AI Professional
* CompTIA CySA+, Security+, PenTest+, Network Security Professional, Security Analytics
* CCNA
* ITIL Foundation

Given this background, I’m considering between:

1. Terraform Associate Certification (infrastructure as code)
2. Kubernetes Certification (container orchestration)
3. Azure Cloud Architect Certification
4. something else? please mention in comments

My goal is to solidify my cloud skills and eventually specialize in telecom cloud or cloud security.

For those experienced in the field, which certification would provide the best return on investment, credibility, and market demand? Also, is there a preferred order to take these certifications for someone starting out? Any insights or personal experience would be greatly appreciated!

Thanks in advance!

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

Kodecloud subscription vs Udemy course for Kubernetes, docker etc

I know there are a lot of courses available, even for free but I have not been able to self study or learn something since quite some time. I believe something like might help and I have heard they have good kubernetes resources. Kodecloud pro is very costly but it gives the benefit of playgrounds so I don't have to setup everything for which i often procrastinate.

So is kodeckoud pro worth it or should I stick to their udemy course?

https://redd.it/1mzi2lj
@r_devops

Any good offline-first alternatives to Postman?

I’ve been hitting a wall with API clients lately. Most of them (Postman, Insomnia, etc.) really push cloud sync and accounts, but sometimes I just want a tool that works locally without sending data anywhere.

Things I’ve found so far:

Bruno → open source, collections saved as plain files. Works great with Git.

Hurl → totally scriptable, stores everything in text format.

Insomnium → fork of Insomnia before it went closed-source.

Apidog → supports offline debugging mode, which helps if you want something modern but not cloud-locked.


Do you think offline-first clients are underrated? Or is cloud sync just too convenient to give up?


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

Anyone solved ECS-to-Loki logging drop after EKS ASG scale-down

I’ve just deployed Loki to my EKS cluster and it’s working great overall.

I’m sending logs from a standalone ECS cluster to Loki via an NLB. The issue: in my DEV/QA environments, the EKS worker node ASGs scale down to 0 after hours and on weekends. The ECS tasks themselves keep running fine, but when the EKS nodes scale back up, those ECS tasks stop sending logs to Loki until I restart them.

Has anyone run into this before? Is there a way to fix it without restarting the ECS tasks?

https://redd.it/1mzlph0
@r_devops

Course after CS bachelor degree

"The ultimate DevOps Bootcamp course for all your DevOps learning" Is a good udemy course from Kodekloud after CS bachelor degree to get Junior position? Can i use it in my resume?

https://redd.it/1mzl7tc
@r_devops

AI/ML Learning path and Conferences

Hi all,

Im experienced devops engineer, as most of the people in this group. My current daily basics are cloud services, docker, kubernetes, CI/CD , automation and a bit etc.

Now DevOps folks are starting to join AI/ML projects and i see lot of work coming from DevOps teams: MCP, gpu sharding, schedule k8s data workloads, inference servers and more to come.

I wonder abouy the best learning path for DevOps to join the AI world ( i dont mean about creating AI tools for DevOps, i mean to get into the AI workd and collaborate with Data Teams).

Also conferences for us that will mean it,

On example is this i found in another reddit post: https://aws.plainenglish.io/ai-roadmap-for-devops-ce6e8f8a8d37



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

How to reduce noise in OpenTelemetry?

A practical guide to reducing telemetry noise with OpenTelemetry—cutting cost & alert fatigue while surfacing the 5% of data that drives 95% of incident resolution.



https://oneuptime.com/blog/post/2025-08-25-how-to-reduce-noise-in-opentelemetry/view




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

CKA vs CKAD Kodekloud course for learning Kubernetes as a intermediate?

Hi, I have been using Kubernetes since about a year. Mostly basics, debugging pods, logs, describe, know basic architecture.

Now, i want to learn kubernetes properly to take my skills to the next level. Which course should I take?
CKA vs CKAD?

https://redd.it/1mzpq8b
@r_devops

Question for CXOs & Leaders: How do you measure the business impact of DevOps?

I was talking with some tech leaders recently, and one theme kept coming up over and over again-DevOps is no longer just about automation and pipelines, it’s about driving business outcomes.

Some key challenges I keep hearing from CXOs:

* How do you show ROI of DevOps investments to the board?
* How do you balance speed, security, and cost while scaling?
* What metrics actually prove DevOps success in business terms?

👉 I’d love to hear from the community - how are you addressing these challenges in your organizations?

To make this conversation more interactive, we have also created a dedicated WhatsApp group where CXOs, managers, and practitioners exchange ideas, share case studies, and discuss leadership-focused DevOps problems.

If you're interested, DM me for the invite link (this group is for knowledge sharing only, not spam).

Looking forward to your feedback here too! 🚀



https://redd.it/1mzk7nx
@r_devops

18x cost blow-up from retry storm + malformed JSON

We just ate $400 in 48 hours from a silent bug.
One request that should’ve cost $0.43 ballooned to $7.81 because a recursive JSON object turned into a 3.2MB payload and got shoved straight into the LLM as “context.”

Monitoring didn’t help — HTTP 200s everywhere, token usage looked high but plausible, cost alerts lagged by 6+ hours, and we had no payload size checks.

Fixes that actually worked:

100KB hard limit at API boundary
Per-request cost tracking with a $3 circuit breaker
Schema validation in CI to catch circular refs
Dedupe script (jq … | awk '$1 > 2')

Results: 91% fewer duplicate requests, and caught 2 more cost bombs (\~$300) before they hit the bill.

Anyone else validating payloads before they hit expensive APIs?

https://redd.it/1mzt9hz
@r_devops

Can I install a cloud like environment in a laptop?

We develop enterprise applications, in azure cloud. But basically the services we use are db, web server, some functions and the service bus. The users and the workload are ridiculous, like in the range of hundred of users, nor even thousands.

The thing is that we deploy everything to cloud, for every environments. Some applications have 5/10 environments for dev only. Multiply by 10 or so integrated applications.

All based on terraform and kubernetes
Wouldn't be possible to setup few local PCs and deploy there? With docker or openshift?

Would it make sense moneywise, including the effort to maintain that? At least for development?

https://redd.it/1mztv7f
@r_devops