Problem Downloading FHO Clips

I’m having trouble downloading input data for the FHO benchmark task; I successfully downloaded the annotations, but I tried downloading clips, full_scale, and slowfast8x8_r101_k400 all with similar issues.

I run the following in my terminal:

ego4d --output_directory="./ego3d_data" --datasets clips --benchmarks FHO

Then it fails after a few minutes with some variation of the following error messages:

urllib3.exceptions.NewConnectionError: <botocore.awsrequest.AWSHTTPSConnection object at 0x12936d540>: Failed to establish a new connection: [Errno 8] nodename nor servname provided, or not known
botocore.exceptions.EndpointConnectionError: Could not connect to the endpoint URL: ""

If you could help me figure out how to trouble shoot this issue, I would greatly appreciate it!

Thank you so much!

Additional Details:

  • Hardware: 2019 Macbook Pro
  • Chip: M1 Max
  • OS: Ventura
  • Python version: 3.10.8

Hey Wesley,

Thanks for your question!

I’m able to run the same command on my machine’s terminal and it downloads the clips as expected. I’m running this on a 2021 M1 Macbook Pro on Ventura.

Can you share your default region name, then try the following?

  1. Run aws configure and ensure the correct Access ID and Access Key are set
  2. Run aws configure and set the “Default region name” to us-west-1
  3. Verify none of your firewall or network settings block access to s3 or this endpoint

Let us know how it goes!
- Devansh

Hi Devansh,

I ran aws configure and double checked the AccessID and Access Key; I also refreshed my credentials and tried again. I also tried changing the default region name to us-west-1 from the blank default suggested in the docs. I’m still getting the same errors.

What network or firewall settings would block s3? I have the default firewall on normally, and I tried turning off the firewall, but still had the same errors.

Also, something odd I noticed is that the progress bar says the total download size is .61M despite a 70.8 GB download:

Any other suggestions?