How one team deployed Apache Storm on AWS EKS without an official Helm chart

·30 views

A development team set out to run Apache Storm on a non-production Amazon EKS cluster, despite no official Helm chart or Kubernetes migration guide existing for the project. Rather than relying on community charts, they built a custom local Helm chart to maintain full control over templates and configurations. The rollout was structured in five phases, starting with ZooKeeper and Nimbus to validate the control plane, then progressively adding Supervisors, the Storm UI, and private ECR images. Early failures were silent rather than dramatic — missing StorageClass definitions left PersistentVolumeClaims stuck in a Pending state, which cascaded into Nimbus never starting and Supervisor pods looping indefinitely. The team noted that a pre-rollout risk list they had written proved more accurate than their original deployment plan, highlighting the value of anticipating failure modes when pioneering untested infrastructure patterns.

Read the full story at DEV Community

This is an AI-generated summary. ShortSingh links to the original source for the complete article.

Discussion (0)

KDE turns 30 and someone's brought an AI-native desktop proposal

Article URL: https://www.theregister.com/software/2026/09/18/kde-turns-30-and-someones-brought-an-ai-native-desktop-proposal/5297282 Comments URL: https://news.ycombinator.com/item?id=49773439 Points: 5 # Comments: 0

0 comments Read more at Hacker News

ProgrammingDEV Community ·

AI Tool Mines GitHub PR Comments to Extract Team Code-Review Rules, Exposes Early Flaws

An AI agent called Instinct Bot, built by developer Ofer Shapira, created an open-source CLI tool named PR Rulebook that scans merged GitHub pull requests to identify recurring, unwritten code-review conventions. The tool analyzed 45 human review comments from 15 merged pull requests in the astral-sh/ruff repository, generating two candidate rules — both of which proved unreliable. A key flaw was discovered: the tool mistook repeated comments within a single PR for team-wide conventions, inflating confidence scores without cross-PR evidence. The developers have since patched the clustering logic to require evidence from at least two distinct pull requests and improved text normalization. The tool remains in early development, with its npm package unpublished, and the team is seeking five public repositories with active human PR review for a pilot run.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Open-Source Spring Boot Starter Prevents Duplicate API Requests via Idempotency Keys

A developer has released an open-source library called Spring Boot Idempotency Starter, designed to prevent duplicate processing when clients retry failed API requests. The starter introduces an @Idempotent annotation that developers can apply to endpoints, along with an Idempotency-Key request header that clients use to tag each unique operation. When a retry is detected with the same key and request body, the server replays the cached response instead of re-executing the controller logic; if the key is reused with a different request body, a 409 Conflict error is returned. The library supports both in-memory and Redis storage backends, configurable expiry periods, and Spring Boot auto-configuration, requiring Java 17+, Spring Boot 3.x, and Maven 3.8+. The author cautions that the tool is not a guarantee of exactly-once execution and recommends pairing it with database constraints and business-level safeguards for critical operations.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Developer Tests API Rate Limiting via WSO2 Bijira, Finds Clock-Window Timing Is Critical

A developer working on a hotel booking backend discovered that his GET endpoint had no rate limiting, forcing him to disable it entirely when exposed publicly for testing. To explore managed rate limiting, he published a read-only endpoint through Bijira, WSO2's API management platform, configuring a 5-requests-per-60-seconds limit via a simple console form with no custom code. Initial tests failed to trigger rejections because two requests hung for 60 seconds each, pushing subsequent requests into new clock-minute windows where the counter had already reset. After switching to a fixed 200ms firing schedule starting just after a new minute began, all 10 requests landed within the same window, yielding exactly 5 approvals and 5 HTTP 429 rejections as configured. The experiment highlighted that rate limit behavior depends heavily on clock-aligned windows and request timing, not just the gap between individual requests.

0 comments Read more at DEV Community