Sushruth

Annoy the world into letting you succeed

@SahilExec promise.race usually takes multiple async reqs and picks the first one that finishes executing. it doesnt stop the rest of the processes So here the fetchUserData would keep running. The correct approach is to use abortcontroller. That would cleanly kill the underlying request.

A junior dev wrote this to handle timeouts: What's the critical flaw and what keeps running after the timeout fires?

@codevsdev DevOps & Cloud All those AI models need to be deployed and all those K8s and infra need to be maintained. its complex and critical enough that theres always a human in the loop no matter how much automation is thrown at it

if you could master one skill instantly, what would you choose? 💻⚡ a) DSA & Problem Solving b) System Design c) AI/ML d) Full-Stack Development e) DevOps & Cloud

@deepp2108 Eyyyy ya boi on a roll Day 5 and already deploying. Crazy work, gg 💪

Day 5 of Showing Up Every Day >Spent most of the day working on a deployment >Fixed a few bugs along the way Deployment took longer than expected That's all for today happy learning!

Since its specifically happening at peak traffic Id look for the bottleneck service first not a general failure. The flow looks synchronous so one slow service stalls everyone upstream and the timeouts cascade. Id check the metrics to find which one degrades first under load. Usual suspects are db connection pool exhaustion or a downstream like payment that cant keep up. Also would check if circuit breakers tripped which tells us where the stall started.

This system starts timing out during peak traffic. What would you investigate first?

Domain expertise. Biggest mistake is not building domain expertise Everyone is chasing generic full stack skills but the generic software dev era is basically over The ppl who survive are the ones who go deep in one niche and actually understand the domain not just the framework.

What's the biggest mistake the tech industry is making right now?

Hahaa I work with card txn processing so this is right up my alley :D There are multiple checks here. First and most important is the luhn algo on entire PAN which is just a checksum so it catches typos and impossible numbers instantly Theres also the BIN(first 6-8 digits of PAN) that maps to networks like MC visa etc.If the BIN isnt identifiable the network either doesnt exist or isnt registered with this gateway These are just the checks that happen without network calls. Post that comes the actual card validation that involves multiple cryptography and person details validations. Its quite interesting what happens behind modern card transactions.

Interviewer: you type a 16-digit card number and the form instantly says invalid. there are billions of possible combinations, how does that check happen so fast ?

@livingdevops I agree. Linux should have been a mandatory course in Compsci to begin with(It isn't in India)

Linux-inspired Docker Linux-inspired Kubernetes (Borg existed because of Linux). Kubernetes heavily shaped modern platform engineering. Yet many “cloud-native” engineers still treat Linux like an afterthought. That’s like trying to become a Formula 1 driver without understanding engines, gearboxes, or aerodynamics. Because underneath all the shiny tooling, everything still runs on Linux. > > Containers = namespaces + cgroups + seccomp + capabilities > > Kubernetes = a scheduler for Linux workloads > > Most production nodes = Linux > > Networking, storage, security, observability = Linux primitives And when production breaks at 3 AM, you usually are not debugging Helm charts. You are debugging: > > cgroup v2 memory pressure > > mount propagation issues > > nftables or eBPF rules > > systemd slice limits > > seccomp or AppArmor denials The engineers who reach Staff+ and Principal levels usually have one thing in common: They are extremely comfortable on a raw Linux machine. Not because Linux is trendy. Because the entire cloud-native stack is built on top of it. Skip the fundamentals and Kubernetes feels like magic. Master Linux and the whole stack becomes understandable.

@javarevisited This should be a join not the comma syntax Comma version works but its riskier and not standard. Also in prod its generally better to avoid SELECT *

Interviewer: What’s wrong with this query? SELECT * FROM orders o, customers c WHERE o.customer_id = c.id;

@devXritesh B is scalable since category is a query param we can stack additional params on top in a generic way without worrying abt new routes for filtering specific products. Even if we pick A it would eventually evolve back into B

Backend Interview Question: Which API design scales better? What's your choice?

Single-threaded is not always a problem, sometimes it's useful In redis' case, it is useful because redis can avoid handling multi-threading synchronisation and other issues that would need to be handled in a typical multi-threaded setup Basically single thread means no locks, no context switching, no contention etc. its in memory so cpu never waits on disk and an event loop handles thousands of connections without blocking. Usually from what I see the main issue that always becomes the primary bottleneck is network infra that surrounds redis not redis itself.

A question that is being asked a LOT in interview these days : Why Redis is single-threaded but still faster than most multi-threaded databases?

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees. The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance. Access to all other Claude models is not affected. We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible. Read our full statement: anthropic.com/news/fable-myt…

@AnthropicAI wow. just wow. I havent even fully tried it out yet T_T

@devXritesh Good morning! Ive been inconsistent as well due to just being busy and yes, life being life. But I am back! Let's get back to consistency together 💪

Good morning and Happy Saturday ☀️ I've been inconsistent over the last 2–3 weeks. Life happens. But starting today, I'm getting back to what I enjoy most: ➜ Backend Development ➜ System Design ➜ Software Engineering ➜ AI & Building Products Expect more interview questions, real-world engineering lessons, and production insights. Let's make the next few months count. Thanks for being here.

@deepp2108 This guy?

Ignoring the elephant in my mind

@theEquinoxDev @deepp2108 Flex

@deepp2108 mai toh unemployed hu

@ashoKumar89 I agree. Last year it was def worse but lately its been getting better as long as we dont have a full on design system to be followed and pixel-specific UI specs

Used Grok Build to generate a Next.js frontend. Honestly… the iteration speed is insane now.

@javarevisited protected branches and PR reviews exist for this reason, the process feels slow until the day it saves you

Tech Lead tells you: “Force push directly to main, nobody will notice.” Will you do it?

Well tbh that is easy to say, but quite often there are many cases where simplicity is simply not an option ryt. More like, the goal is minimum complexity that solves the actual problem, CAP theorem alone guarantees you'll face tradeoffs no clean architecture escapes. Over engg is real but under engg has a body count too

The best architecture is the one you can explain to a junior engineer in 5 minutes. If your system design requires a PhD to understand, you've over-engineered it. Simplicity scales better than complexity ever will.

Cuz nginx is a single process. We'd still have to manage and keep alive, dedicated load balancers like AWS ALB are managed redundant infra with builtin failover, health checks, SSL termination and autoscaling hooks. We ideally don't want our load balancer to be another thing that can go down at scale.

Interviewer: If Nginx can load balance traffic, why do companies still use dedicated load balancers? Interviewer:

@SumitM_X Means we're not on a branch, commits here get GCed git branch recovery-branch then git checkout recovery-branch to save your work If no commits yet just checkout the branch you meant to be on

You run : git commit and see below error : fatal: You are in 'detached HEAD' state. What are your next steps ?