PKPrashantkumar Khatriinrootcause.hashnode.dev·1d ago · 8 min readAnatomy of a score.yaml: How Resolution Actually WorksScore is marking two years in the CNCF Sandbox this July, though the project itself predates that. Most of what I've read about it walks through a score.yaml, runs a command, and shows an app running 00
FFerinferztyle.me·2d ago · 18 min readKubernetes resource requests and limits explained: scheduling, throttling, and OOMKillThis is part of the Platform engineering with Go series: a growing collection of posts on Kubernetes, Go tooling, and infrastructure automation. View all posts in the series The 3am incident nobody t00
GSGeorge Simsindowntherabithole.dev·3d ago · 7 min readHow a Process Really WorksAs a DevOps/SRE/Platform Engineer you see many varieties of processes in the wild; Microservices writing to a database, a CI pipeline which runs linters and unit tests, the Docker ecosystem itself is 00
MAMateen Anjuminmateenali66.hashnode.dev·2d ago · 11 min readAI on Kubernetes in Production: The Complete Accountability Stack for GPU WorkloadsExecutive Summary Kubernetes 1.32 introduced Dynamic Resource Allocation (DRA) as GA, replacing device plugins for structured GPU allocation GPU node taints and namespace ResourceQuotas are the minim00
MAMateen Anjuminmateenali66.hashnode.dev·2d ago · 6 min readStop Writing Helm Charts as Your IDP: Build Self-Service Platform APIs with Crossplane and kroExecutive Summary Problem: Helm charts expose too much Kubernetes complexity to developers, creating a bottleneck at the platform team Solution: Crossplane (CNCF graduated) + kro (CNCF sandbox) enabl00
RTRAMIZ TAKILDARinagentic-devops.hashnode.dev·6d ago · 17 min readAgentic AI for DevOps EngineersIntroduction Why This Topic Matters You've probably used ChatGPT to write code, debug errors, or explain complex concepts. It's impressive, but have you ever noticed its limitations? It can't execute00
SBSamanta Bindrainarchitecturedev.hashnode.dev·Jun 27 · 8 min readResilience Patterns Break When They Run TogetherResilience patterns get taught individually. Circuit breaker, bulkhead, and retry with exponential backoff. Diagrams with clean failure paths and tidy recovery arrows. What happens when a circuit open00
SBSamanta Bindrainarchitecturedev.hashnode.dev·Jun 27 · 8 min readDead Letter Queues in KafkaA DLQ without replay is a graveyard. A DLQ without classification is noise. A DLQ without age-based alerting is a silent data loss problem you won't notice for days. Every Kafka consumer will eventua00
SBSamanta Bindrainarchitecturedev.hashnode.dev·Jun 27 · 7 min readDistributed Rate LimitingMost discussions around rate limiting start with algorithms — Fixed Window, Sliding Window, Token Bucket, Leaky Bucket. After building and operating a distributed rate limiting layer, one thing become00
SBSamanta Bindrainarchitecturedev.hashnode.dev·Jun 27 · 8 min readExactly-Once Processing in ProductionWhat this article covers: how idempotency, the outbox pattern, and saga orchestration work together as a single coherent system — not three independent patterns bolted onto each other. Exactly-once p00