Welcome to the first 2026 edition of our monthly Azure Container Apps (ACA) update!

Networking: consumption vs workload‑profile environments

A new community article clarified the difference between consumption and workload‑profile environments. In a consumption environment the platform handles scale and node selection but you cannot control outbound routing – there is no support for user‑defined routes (UDRs), NAT gateways or custom firewalls (Container Apps Environment Networking). The guide explains when to choose each option and includes a Bicep snippet for creating a consumption environment.

Exporting telemetry via OpenTelemetry

Icefire's notebook provides a practical guide for exporting logs, metrics and traces from ACA to external systems. It notes that ACA includes a managed OpenTelemetry agent, enabling you to send telemetry to services like Datadog or an Aspire dashboard without running your own collector (How to Export Logs, Metrics, and Traces from Azure Container Apps with OpenTelemetry). The article walks through creating a .NET 10 application, adding OpenTelemetry packages and deploying to ACA, then configuring the environment with az containerapp env telemetry commands to send OTLP data to your chosen backend.

Hosting remote MCP servers

A detailed tutorial from ATA Learning explains how to host Model Context Protocol (MCP) servers on Azure. It highlights ACA as the most flexible compute option thanks to its serverless scaling, idle billing and support for HTTP/1.1 and HTTP/2 connections (A Practical Guide to Hosting and Managing Remote MCP Servers on Azure). To expose an MCP server to agents, you need to set ingress to external and choose the appropriate transport (HTTP/1.1 or HTTP/2) when creating the container app (Host remote MCP servers in Azure Container Apps). The article shows how to configure managed identities, environment variables and monitoring for a production‑ready MCP server.

Azure Developer CLI (azd) January release

The January 2026 release of the Azure Developer CLI introduced several improvements for ACA deployments:

• Bug fixes for ACA deployments: azd now correctly resolves the infrastructure path when using layered provisioning and honours the module setting when creating new revisions (Azure Developer CLI (azd) - January 2026). Error messages were improved for non‑revision deployments and environment variables now sync correctly with remote CI/CD pipelines (GitHub Releases). The release also removed Azure Spring Apps support, recommending Container Apps or App Service as alternatives.
• New templates and samples: The January release added more than a dozen templates for ACA, including MCP servers with Copilot Studio agents, LangChain.js and LlamaIndex orchestration, FastAPI/PostgreSQL apps, multi‑agent support triage systems and image search powered by Azure AI Search (azd Changelog). These templates help developers get started quickly with multi‑agent workloads and AI‑augmented web APIs.

Deploying your own 24/7 AI assistant (MoltBot)

A Tech Community post demonstrates how to deploy the MoltBot AI assistant on ACA in about 30 minutes. The author chose ACA because it provides strong security and compliance features: managed identity and built‑in secrets management avoid storing credentials in config files; VNet integration and private endpoints limit exposure; containers run in Hyper‑V isolation; TLS certificates are automatic; and the platform scales automatically (Deploy MoltBot to Azure Container Apps). ACA's integrated monitoring also surfaces logs and metrics in Application Insights (GitHub - MoltBot Azure Container Apps). The sample walks through creating a Discord bot, cloning the MoltBot repo and running azd up to deploy the bot. Operating costs are estimated at US $40– per month.

Observability and ticketing with SRE Agent

Dan Chelupati's blog shows how to extend the Azure SRE Agent by hosting Grafana and Atlassian MCP servers on ACA. Because some MCP servers are only available as local binaries, he deployed them as ACA instances and exposed the /mcp endpoints via public ingress (How SRE Agent Pulls Logs from Grafana and Creates Jira Tickets). The SRE agent then connects to these endpoints to query Loki logs, identify rate‑limit errors and automatically create Jira tickets during incidents (GitHub - grocery-sre-demo). This approach demonstrates ACA's flexibility in hosting third‑party MCP tools and integrating them with agentic workflows.

Running Python workloads with Orchestra

Orchestra, a data orchestration platform, published a guide for running Python scripts on Azure using Container App Jobs. It explains that you need an ACA environment with public internet access to pull agent images from a registry:

• Orchestra Azure Container App Jobs Docs
• Get started with Python on Azure | Orchestra
• Orchestra Terraform Module

Thanks for reading! We hope this update helps you stay up‑to‑date with Azure Container Apps. See you next month.