Roheet Mane

Cloud & DevOps Engineer | CI/CD • Automation • Cybersecurity | Designing scalable, secure & cost-efficient cloud infrastructure

[Roheet Mane @Roheet_Mane_in]

Kubernetes is great at scaling systems.

But it won’t fix inefficiency.

If your app is wasteful,
K8s will just scale the waste.

Optimize first.
Then scale.

#Kubernetes #DevOps #CloudNative

[clovis @clovisdsdo]

Kubernetes Networking -- part 1/10

1 - The Only Thing You Need to understand:
A request is just moving through layers
- DNS ---> finds where to go
- LoadBalancer ---> receives traffic
- Ingress ---> routes HTTP requests
- Service ---> forwards using IP + Port
- Pod ---> runs your application
That’s it.

2 - Each step handles a different layer of the OSI model:
-Layer 7 ---> HTTP (Ingress, App)
- Layer 4 ---> TCP (Service, LB forwarding)
- Layer 3 ---> IP routing (inside cluster)
k8s make network layered and abstracted

3 - Next post:
We start at the Layer 7 (HTTP) and how requests are actually routed.

[Uday👨‍💻 @uday_devops]

Linux file permissions 📚

[Yoshik @AskYoshik]

The skills that got you hired in 2026 won't matter by 2030.

What becomes less valuable:

- Writing Docker files manually
- Knowing specific CI/CD tools
- Memorising cloud service names
- YAML engineering skills

What becomes irreplaceable:

- Understanding TCP behaviour under packet loss
- Reading system calls and kernel traces
- Debugging race conditions in distributed state
- Knowing why eventual consistency breaks

Because when every tool generates code, the only advantage is understanding what the generated system actually does at runtime

[Nandkishor @devops_nk]

ArgoCD is evolving and this release actually solves real pain.

After working with GitOps setups across multiple environments, one thing is clear:

- Day-2 operations are always harder than Day-1 setup.

The upcoming ArgoCD 3.4 release focuses exactly on that.

Here are 4 updates that genuinely matter:

🛑 Pause at Cluster Level
- Finally, a simple way to stop reconciliation during incidents.
- No more disabling apps one by one.

🧩 Helm Value File Flexibility
- Wildcard support = less config mess, fewer PRs, cleaner structure.

🔍 Better UI Filtering
- Filtering by annotations & status makes the UI actually usable at scale.

⚡ Reduced DNS Noise
- Less background chatter = better performance + cost savings.

Small changes. Big operational impact.

Especially for teams:
• Running multi-cluster GitOps
• Handling production incidents
• Managing large-scale platforms

If you're using ArgoCD in production Which feature would help you the most ?

[Roheet Mane @Roheet_Mane_in]

How it’s sold vs. how it’s lived 😅
“Auto everything” in the daylight…
“Where are the logs?!” at 3AM 🚨🔥
#DevOps #RealityCheck #OnCallLife

[Roheet Mane @Roheet_Mane_in]

Kubernetes doesn’t make systems resilient.

It gives you the tools.

What matters: • resource limits
• probes
• autoscaling
• observability

Bad architecture?
K8s exposes it faster.

#Kubernetes #DevOps #CloudNative #SRE

[Roheet Mane @Roheet_Mane_in]

Downtime rarely starts with a crash.

It starts with small, ignored signals: • slow queries
• retry storms
• queue buildup

By the time it’s “down” —
it’s already too late.

Monitor early.
Act sooner.

#DevOps #SRE #Reliability #Tech

[Roheet Mane @Roheet_Mane_in]

Every incident is a design document you ignored.

• alerts tell you what broke
• logs tell you why
• metrics tell you when
• traces tell you where

If you’re missing one — you’re guessing.

#DevOps #SRE #Observability #Tech

[Roheet Mane @Roheet_Mane_in]

Production rule:

If you can’t detect it,
you can’t fix it.

If you can’t fix it fast,
you don’t understand it.

Observability isn’t optional —
it’s your first line of defense.

#DevOps #SRE #Observability #Tech

[Roheet Mane @Roheet_Mane_in]

If your system needs manual intervention to recover,
it’s not production-ready.

Design for:

automated rollback

self-healing

observability-first debugging

Uptime isn’t luck.
It’s architecture.

#DevOps #SRE #CloudEngineering

[Roheet Mane @Roheet_Mane_in]

Good engineers build systems that work.

Great engineers build systems that fail gracefully.

Because failure isn’t an edge case in tech —
it’s the default.

Design for it.

#DevOps #SRE #Engineering

[Roheet Mane @Roheet_Mane_in]

Most companies think DevOps is about speed.

It’s not.

It’s about reducing the cost of failure.

Fast deployments mean nothing if rollback, observability, and ownership are weak.

#DevOps #SRE #Engineering