SkyPilot - Issue launching V100 on AWS with Deep Learning AMI GPU PyTorch 2.0.1

SkyPilot

skypilot-org.slack.com

skypilot-users

Issue launching V100 on AWS with Deep Learning AMI GPU PyTorch 2.0.1

I'm encountering an issue when trying to launch a V100 on AWS with the 'Deep Learning AMI GPU PyTorch 2.0.1'. The instance starts momentarily and then is immediately stopped. How can I resolve this?

Jason Krone

Asked on Oct 07, 2023

The issue is caused by SkyPilot hard-coding the SSH user as 'ubuntu', while the image expects logins from 'ec2-user'. To resolve this, you can temporarily hardcode the 'ec2-user' as the SSH user.

Oct 07, 2023Edited by