lambdash-efs

This is a fork of lambdash by Eric Hammond, an AWS Lambda function that simply runs a shell command inside the Lambda runtime and returns its standard output & error; accompanied by a local command-line invocation helper.

We've modernized it to deploy using Terraform, and to make the Lambda function mount an existing EFS file system at /mnt/efs. This provides a serverless facility to browse & manage EFS contents, at negligible cost whenever needed; filling a gap that exists while AWS Console lacks a browser for EFS contents, CloudShell can't mount EFS, etc.

How to

Requirements:

• terminal session with AWS CLI configured (administrator/poweruser credentials and default region)
• git, terraform, python3, boto3
• an EFS file system, and specifically:
  • EFS Access Point (fsap-xxxx)
  • VPC subnet (subnet-xxxx) that can reach the EFS
  • Security Group (sg-xxxx) that can reach the EFS

Deployment:

git clone https://github.com/miniwdl-ext/lambdash-efs.git
cd lambdash-efs
terraform init -upgrade
terraform apply -var=fsap=fsap-xxxx -var=subnet=subnet-xxxx -var=sg=sg-xxxx

Example session:

$ ./lambdash uname -a
Linux 169.254.195.245 4.14.252-207.481.amzn2.x86_64 #1 SMP Wed Oct 27 20:57:19 UTC 2021 x86_64 x86_64 x86_64 GNU/Linux
$ ./lambdash bash -c 'echo Hello, world! > /mnt/efs/hello.txt'
$ ./lambdash ls /mnt/efs
hello.txt
$ ./lambdash cat /mnt/efs/hello.txt
Hello, world!
$ ./lambdash rm /mnt/efs/hello.txt

And naturally you may copy the lambdash script somewhere in your PATH.

Limits

The Lambda function is configured with 256 MiB memory and 60-second timeout. The command's stdout & stderr are truncated at 64 MiB each.

Alternatives

• Simple File Manager for Amazon EFS serverless web app (not really so simple...)
• SageMaker Studio managed JupyterHub accessed via AWS Console, attaching a built-in EFS (can't bring your own EFS)
• Cloud Commander web app to deploy on EC2 or Fargate, mounting your EFS (similarly JupyterHub, code-server)
• SSH into an EC2 or Fargate server mounting your EFS