AWS Lambda Managed Instances: Setting Min/Max to Zero Deactivates Functions, Not Idles Them

·29 views

AWS Lambda Managed Instances (LMI), introduced at re:Invent 2025, places Lambda functions on EC2 capacity while retaining Lambda's programming model, but its scaling behavior differs sharply from standard Lambda. When both MinExecutionEnvironments and MaxExecutionEnvironments are set to 0, LMI does not idle the function awaiting the next request — it fully deactivates the function version and terminates all backing EC2 instances. Any invocation against a deactivated version returns a hard failure with no cold start, no queuing, and no automatic retry or reactivation. Reactivation requires a manual intervention, such as pushing a new non-zero scaling configuration via the console, the PutFunctionScalingConfig API, or a scheduled action. This behavior is fundamentally different from standard Lambda concurrency throttling and represents a critical operational distinction for engineers running non-continuous workloads on LMI.

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)

Scaffoldr Proposes Product-Based Project Starters Over Stack-Based Scaffolding

The author argues that existing tools like create-next-app, create-t3-app, and SaaS starter kits organize project setup around technology stacks or fixed product templates. Modular scaffolders ask which technologies to use, while starter kits lock developers into someone else's choices, which can drift over time. The proposed Scaffoldr tool would instead let developers specify a product type, such as ticketing, with swappable modules for payments, databases, auth, and email. The author notes that African payment providers like Paystack are often overlooked, and the project is not yet built, with maintenance and developer interest remaining open challenges.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Yonto Player moves its source contract into JavaScript plugins on QuickJS

The Yonto Player TV app connects to media servers and folders through a single ContentSourceAdapter interface. Its source contract began as an HTTP API backed by a hosted service, but that required someone to run and update a middle layer whenever a media server changed its fields. The developer kept the contract's JSON shapes as the single definition and made transport an implementation detail, allowing sources to run as JavaScript plugins inside the app. Each plugin is one file whose manifest, declaring its identity and permitted hosts, is read as plain text, and it exposes required functions such as search and getMediaList. Plugins reach the outside world only through a restricted global object, which limits them to requests, configuration, a small cache, logging and parsing helpers.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

OAB rules set guardrails on AI use in legal practice, keeping lawyers responsible

The article explains guardrails for AI in law: technical and ethical limits on using AI for tasks like drafting petitions, analyzing evidence, and giving procedural guidance. It notes that OAB Resolution 05/2023 bars attributing professional responsibility to AI systems and requires lawyers to keep full control, critically review outputs, and accept accountability. The resolution also calls for auditable, transparent, bias-free tools, and platforms must disclose when content is machine-generated. Practical guardrails include blocking drafts unsupported by validated data, flagging conflicts of interest, and logging all AI interactions. Ultimately, the lawyer remains responsible for any AI-assisted work.

0 comments Read more at DEV Community

ProgrammingDEV Community ·

Guardrails para Claude, GPT e modelos próprios: o que são e por que importam

Guardrails são camadas técnicas e políticas que limitam saídas de IA consideradas inseguras, antiéticas ou ilegais, combinando filtros de entrada e saída, classificadores de conteúdo, regras de negação e treinamento por feedback humano. No Brasil não há lei federal específica sobre o tema, mas a Estratégia Nacional de Inteligência Artificial recomenda controles éticos e técnicos, e setores regulados devem alinhar seus sistemas às normas de seus órgãos. O Código de Defesa do Consumidor e a LGPD reforçam a responsabilidade dos fornecedores por danos causados por saídas automatizadas. Especialistas destacam que guardrails são sistemas de governança contínua, exigindo testes, atualização e supervisão humana, e não apenas listas de termos proibidos.

0 comments Read more at DEV Community