TLDR DevOps 2026-09-02

Technical leaders from Intel, IBM, Uber and more will gather in San Jose this September to talk search, observability, and AI workloads, and to connect with peers solving the same problems.

Get your ticket for keynotes, hands-on sessions, and more:

• Leave with real production patterns for your team, not just slides
• Build a full OpenTelemetry observability stack or an agentic AI search app in two hands-on Developer Workshops
• Network with peers tackling the same search and observability challenges

See the full schedule and grab a seat

Storage version migration (SVM) is now generally available in Kubernetes v1.37, stable and enabled by default across all clusters running that version. The feature lets cluster administrators migrate stored API resources to a current storage version by creating a declarative StorageVersionMigration object, which the built-in control plane controller then handles automatically. Previously, admins had to rely on manual kubectl scripts or a separate out-of-tree component called kube-storage-version-migrator.

Wasmi 2.0 is roughly 2.2× faster than Wasmi 1.0 across its benchmark suite, largely through a redesigned interpreter that adds threaded-code dispatch, accumulator registers, and lower-overhead instruction handling. The release also preserves fast startup, adds options to reduce binary size, and includes stable fuel metering, deterministic WebAssembly support, and an improved CLI.

AI-driven development accelerates technical debt, making continuous modernization essential in legacy environments. A DIY approach using AWS Transform custom and GitHub Actions automates dependency remediation, documentation, repository-wide transformations, and continual learning, enabling ongoing modernization through CI/CD.

AWS has published a guide for building an automated incident response pipeline on Amazon EKS using a new open source tool called the DevOps Agent Operator. The Operator watches for pod failures like OOMKilled or IP exhaustion, immediately collects pod logs, node-level data, and Kubernetes events, then sends that data to AWS DevOps Agent via a webhook before evidence disappears.

Agents now write 76% of Laurel's PRs. Each one is verified pre-merge in a production-like environment on their own cluster. Only the changed services spin up. Nothing else is duplicated. With Signadot and coding agents, their team of ~60 engineers ships 2,300+ verified PRs per month. Read the case study.

ECC is an open-source engineering harness for coding agents that adds a reusable workflow around planning, testing, implementation, review, verification, memory, and continuous improvement. It ships with specialized agents, skills, hooks, rules, persistent memory, and security scanning, with strongest support for Claude Code plus integration paths for Codex and other agent environments.

OpenClaude is an open-source command-line coding agent that works with multiple AI backends, including OpenAI-compatible APIs, Gemini, GitHub Models, Codex, and Ollama.

Grafana 13.2 expands shared saved queries with search, RBAC, command palette access, and Terraform provisioning, while the new View panel sidebar improves exploration without edit permissions. Additional updates include Git Sync enhancements, a refreshed homepage, secure cloud data source authentication, improved variable editing, and dashboard organization.

Datadog monitored its own AI agents to uncover security lessons, starting with building an inventory of agent components. The effort also traced sensitive data flows and tool calls to identify risks across the agent stack.

OpenTelemetry achieved graduated status within the Cloud Native Computing Foundation in May, making it one of the few projects to reach that tier alongside Kubernetes and Prometheus. Graduated status means the project's contributor health, security standards, governance processes, and adoption have been evaluated at a level suitable for enterprise use at any scale.

Microsoft Azure autoscales when traffic spikes—adjusting capacity automatically so you can skip the scramble. With more global regions than any other cloud provider, you've always got room to grow. Scale with Azure →

Amazon EC2 R9g and R9gd instances are now generally available, powered by AWS Graviton5 processors.

GCP cost optimization becomes difficult when billing data is disconnected from operational telemetry, making cost anomalies harder to attribute and resolve.

Datadog Serverless Monitoring now traces requests, reports enhanced CPU metrics, and profiles code across all Azure Functions hosting plans.