Jetbot ROS2 #4: Continue setup including Xavier

Restoring AWS setup

I tried to setup a gpu instance under AWS, had issues. I thought I would return to my previous work to get a basic instance up. I’m monimally following https://rosdabbler.github.io/notes/2021/10/aws-for-openvpn/

I configured a new elastic IP, now I have 35.81.140.242 I added that as the ip address for ubuaws.caspia.net. ping ubuaws.caspia.net returns this now.

Now I’m going to try to launch an instance. I have an existing security group ‘openvpn’ which I can use. I’ll use the aws-102021 keypair. My .ssh/config has this entry:

Host ubuaws
  HostName ubuaws.caspia.net
  IdentityFile ~/.ssh/aws-102021.pem
  User ubuntu
  StrictHostKeyChecking no

In ‘Choose an Amazon Machine Image (AMI)’ I search for ubuntu. In the quickstart, I see ‘Ubuntu Server 20.04 LTS (HVM), SSD Volume Type - ami-0892d3c7ee96c0bf7 (64-bit x86)” I’ll choose that one. Instance type: t2.micro

‘Configure Instance Details’: Network” mu custom ‘openvpn’ (which is my default). All other entries default. Nothing to do in ‘Add Storage’ and ‘Add Tags’. Under ‘Step 6: Configure Security Group I don’t see the one I was expecting. Why? OK, the ‘Name’ field is not showing, only the “Security group name’ which is default. ID is sg-ea51a2e1. Confusing, but I select that. I launch, selecting aws-102021 keypair. Now how to assign my IP address? I do that in the elastic IP area. now ssh ubuaws connects. Yay. Now I terminate.

Can I follow the same procedure to start a GPU-enabled instance? I’m going to try a g4dn.xlarge instance. It fails with: “You have requested more vCPU capacity than your current vCPU limit of 0 allows for the instance bucket that the specified instance type belongs to. Please visit http://aws.amazon.com/contact-us/ec2-request to request an adjustment to this limit.” I tried a difference type, p2.xlarge. Also failed. So I will make a request.

Trying Google GCP setup

Since AWS is blocking me, I tried Google. I tried to setup a basic instance with a T4 gpu. I got the message:

Operation type [insert] failed with message "The zone 'projects/stalwart-bliss-270019/zones/us-west1-b' does not have enough resources available to fulfill the request. Try a different zone, or try again later." 

OK trying a different region. Same message, but zone is now us-central1-f.

Message does not mention quotas, but googling showed this as an issue. I checked my limits, they are 0 gpus. Boo. I requested an increase.

Nano jetson

I built the docker images on jetson. But when I run I get this error:

kent@nano:~/github/rkent/bdbd2_ws/docker…
(main) $ docker run -it --rm --runtime nvidia bdbd2/jtransformers bash
docker: Error response from daemon: failed to create shim: OCI runtime create failed: container_linux.go:380: starting container process caused: error adding seccomp filter rule for syscall clone3: permission denied: unknown.

Googling I came across https://github.com/dusty-nv/jetson-containers/issues/108 where dusty-nv proposes this fix:

distribution=$(. /etc/os-release;echo $ID$VERSION_ID) \
   && curl -s -L https://nvidia.github.io/nvidia-docker/gpgkey | sudo apt-key add - \
   && curl -s -L https://nvidia.github.io/nvidia-docker/$distribution/nvidia-docker.list | sudo tee /etc/apt/sources.list.d/nvidia-docker.list

sudo apt-get update
sudo apt-get install nvidia-docker2=2.8.0-1

Installation went OK, now trying a reboot. Worked.

That let me test nano with dialogpt. With gpu, speed is 0.92 seconds.