Mimir v2

Highlights

Mimir v2 packages the full local AWS sandbox into one Docker image. The console, API bridge, and AWS-compatible backend now start together, so users can run Mimir without cloning the repo, building from source, or wiring separate services.

What's New

• Bundled AWS-compatible backend served on localhost:4566.
• One-image run flow with tanujsoni027/mimir-aws:latest.
• Multi-arch image support for linux/amd64 and linux/arm64.
• 54 backend services available through the local AWS endpoint.
• 15+ console workflows for common AWS services.
• Container-backed Lambda, EC2, and Glue workflows.
• Browser EC2 terminal for local EC2 containers.
• Real local Glue runtime path for jobs and notebooks using Glue-style Python/Spark code.
• Updated documentation for the v2 run flow, v1/v2 comparison, performance profile, and technical architecture.

Run

docker run -d --name mimir \
  -p 8080:80 -p 4566:4566 \
  -v /var/run/docker.sock:/var/run/docker.sock \
  -v /tmp/mimir-glue:/tmp/mimir-glue \
  tanujsoni027/mimir-aws:latest

Open:

• Console: http://localhost:8080
• AWS endpoint: http://localhost:4566

Default local credentials:

• Access key: mimir
• Secret key: mimir
• Region: us-east-1

Docker Hub

Docker Hub now uses a single current tag:

• tanujsoni027/mimir-aws:latest

Documentation

• Main README now focuses on the one-image product flow.
• docs/architecture.md contains implementation details for contributors.
• docs/v1-readme.md preserves the v1 README for historical reference.

Release Credits

• Tanuj Soni (@tanuj24) - project author and maintainer.

Mimir v1.1

Mimir v1.1 makes the compute services run real workloads: AWS Glue jobs execute unmodified Glue code on the official Glue runtime, AWS Lambda gains a full function console, and the whole stack is easier to keep an eye on.

AWS Glue — real runtime

• Jobs and notebooks run on the official AWS Glue runtime images (Glue 5.0/4.0/3.0/2.0), so unmodified Glue code runs locally with the real awsglue library, a version-matched Spark, and the same preinstalled libraries (boto3/pandas/numpy/pyarrow) as AWS.
• GlueContext, DynamicFrame, the Glue transforms, Job, and getResolvedOptions all work — for both Spark (glueetl) and Python-shell jobs.
• create_dynamic_frame.from_catalog reads from the local Glue Data Catalog; s3:// / s3a:// reads and writes are wired to local S3.
• Notebook run logs stream live, and kernels auto-stop after 120 min idle, resuming transparently on the next cell.

AWS Lambda — full function console

• New per-function detail view: Code, Test, Configuration, Aliases & versions, and Triggers.
• Edit and redeploy code (inline editor or .zip); configure memory, timeout, ephemeral /tmp storage, handler, runtime, environment variables, X-Ray tracing, reserved concurrency, Function URLs, tags, and async-invocation settings.
• Publish versions, manage aliases, invoke a specific version/alias, and wire triggers (event source mappings for SQS / DynamoDB / Kinesis).
• Create flow adds architecture (x86_64 / arm64), ephemeral storage, and more runtimes (Node.js 18/20/22, Python 3.9–3.13, Ruby 3.3, Java 17/21, provided.al2023).

Observability & lifecycle

• A constant bottom resource monitor bar shows host CPU, RAM, and running container count.
• EC2 instance containers are cleaned up on terminate, and orphans from terminated instances are reaped automatically.
• Build and Docker fixes for a smoother docker compose up.

Full Changelog: v1.0...v1.1

Mimir v1.0

Full Changelog: https://github.com/tanuj24/mimir/commits/v1.0