Things I learned while building · 25 posts
The final part of the JVM concurrency series. A four-question framework for choosing between Virtual Thread, Reactor, and Coroutines, plus the unique strengths of each technology and the most common misconceptions.
How virtual thread mount/unmount works under the hood, how it differs from coroutine continuations, why AOP proxies just work, and the synchronized pinning trap with practical fixes.
How Spring wraps suspend fun in mono {} to run it on the Reactor pipeline, and the structural limit where AOP proxies mistake COROUTINE_SUSPENDED for method completion — including why @Transactional + coroutines does not work.
Writing Reactor-style deep flatMap chains as sequential, synchronous-looking code with Kotlin Coroutines. Covers suspend, CPS transformation, state machines, Flow, Channel, and structured concurrency.
Starting from the one-request-one-thread limit of Spring MVC, this post digs into how WebFlux actually works — the Netty event loop architecture, MVC vs WebFlux request handling, and what you must never do on an event loop thread.
From the Pull vs Push model to the four Reactive Streams interfaces, Reactor's Mono/Flux and core operators, and backpressure — building a solid mental model of reactive programming.
From Thread and Runnable to Callable/Future, ExecutorService, and CompletableFuture — tracing how Java's concurrency API evolved toward "stop managing threads yourself."
The difference between processes and threads, concurrency vs parallelism, and why synchronous/asynchronous and blocking/non-blocking are two independent axes — foundational concepts for the JVM concurrency series.
Comparing how the Micrometer (library) approach and the OTel Java Agent (bytecode manipulation) approach work — a look inside premain, the Instrumentation API, and ByteBuddy code injection.
How to propagate context in coroutines, where the thread can change at every suspend point — the structure of CoroutineContext, ThreadContextElement, and bridging ThreadLocal with asContextElement().
Why ThreadLocal-based traceId propagation fails on the WebFlux event loop, and how Reactor Context's bottom-up propagation model solves it — with hands-on examples.
How ThreadLocal works under the hood, how MDC gets traceId into your logs, and the moments Spring MVC actually switches threads — the fundamentals you need before going async.
Why the three pillars of observability need distributed tracing, the history from Zipkin B3 through the OpenTracing/OpenCensus split to the OpenTelemetry merger, and where the Spring ecosystem stands today.
What the three AccessModes really mean, the Deployment + RWO PVC trap, the PV lifecycle and Reclaim Policy — the problems you actually run into when operating PV/PVC.
From the Kubernetes volume concept and Volume types like emptyDir and hostPath, to the PV/PVC architecture that decouples storage from the Pod lifecycle — the first post in the storage series.
From kubectl apply to a running container — tracing step by step how the API Server, Controller Manager, Scheduler, and kubelet each play their part in placing and running a Pod.
Who assigns Pod IPs, and how do Pods talk to each other? Following the packet flow through CNI, veth pairs, the Linux bridge, cross-node routing, and CoreDNS service discovery.
Tracing the full path of an external HTTP request through LB → Ingress Controller → Service → Pod — covering kube-proxy and iptables, conntrack, and troubleshooting commands for every hop.
How to bring CPU, memory, disk, and network metrics together on a single CloudWatch dashboard, then collect server logs and connect alarms to complete your EC2 monitoring setup.
How to collect the memory and disk metrics that EC2 basic monitoring leaves out, using the CloudWatch Agent. Step-by-step coverage from creating the IAM role to installing and configuring the agent, plus its resource overhead on a t3.micro.
How the EC2 monitoring tab relates to CloudWatch, basic vs. detailed monitoring costs, and what the free tier covers — part one of a series on monitoring EC2 server health.
Why Spring Cloud Gateway alone fails DNS resolution on Apple Silicon Macs (Netty's own DNS resolver) and how to fix it — plus a look at native library dependencies and Gradle Configuration concepts.
Enable one-indexed-parameters and @PageableDefault(page = 1) suddenly points at the second page — tracing the cause through the Spring Data source code and tests, and pinning down the correct usage.
How to expand an EBS volume on a full EC2 instance with zero downtime — diagnosing disk usage, emergency cleanup with docker system prune, and applying the new size with growpart and resize2fs.
The full walkthrough of running WordPress on AWS EC2 with Docker Compose and wiring up a domain and HTTPS through Cloudflare — SSH config, Security Groups, and Elastic IP, written for first-time server builders.