Chris2pi Posts - Answer Overflow

Chris2pi

•Created by Chris2pi on 2/12/2025 in #help

newgrp not working in startup script

This might just be a Linux 101 issue but still can't work out why. We have a custom built AMI with Docker installed and want to grant the Coder user access to use Docker. Since the Coder user doesn't exist on the AMI yet I added the commands to the startup_script in resource "coder_agent" The script runs

 newgrp docker

 newgrp docker

and it doesn't appear to work. We run the same command in the terminal and it works. So maybe something strange is happening with the session?

resource "coder_agent" "dev" {
  count          = data.coder_workspace.me.start_count
  arch           = "amd64"
  auth           = "aws-instance-identity"
  os             = "linux"
  startup_script_behavior = "blocking"
  startup_script = <<-EOT
    set -e

    # Add any commands that should be executed at workspace startup (e.g install requirements, start a program, etc) here
    # docker group already exists
    sudo usermod -aG docker coder
    newgrp docker

resource "coder_agent" "dev" {
  count          = data.coder_workspace.me.start_count
  arch           = "amd64"
  auth           = "aws-instance-identity"
  os             = "linux"
  startup_script_behavior = "blocking"
  startup_script = <<-EOT
    set -e

    # Add any commands that should be executed at workspace startup (e.g install requirements, start a program, etc) here
    # docker group already exists
    sudo usermod -aG docker coder
    newgrp docker

9 replies

CCoder.com

•Created by Chris2pi on 1/8/2025 in #help

Persistent AWS EBS/EC2 home directory

Problem—When I click restart in the coder web GUI, it recreates my workspace and causes the EBS volume to be recreated, causing me to lose all of the data. I understand why that is as a terraform recreate is happening. So, before I go down the rabbit hole of making custom scripts, etc., I just wanted to see if there was an out-of-the-box method from the coder. I found this doco - https://coder.com/docs/admin/templates/extending-templates/resource-persistence But it doesn't seem right for what I want. Napkin-style plan for what I think should happen - The user hits restart. Recreation happens, but the EBS is detached from the instance. A new instance spins up and attaches the EBS again. Source code based on this template - https://github.com/coder/coder/blob/main/examples/templates/aws-linux/main.tf

resource "aws_instance" "dev" {
  ami               = data.aws_ami.ubuntu.id
  availability_zone = "${data.coder_parameter.region.value}a"
  instance_type     = data.coder_parameter.instance_type.value
  subnet_id         = "subnet-xxxxxxxxxxxxxxxx"
  associate_public_ip_address = true

  user_data = data.cloudinit_config.user_data.rendered

    # Define the root volume storage
  root_block_device {
    volume_type           = "gp3"  # You can choose gp2, gp3, io1, etc.
    volume_size           = data.coder_parameter.disk_space.value     # Specify the volume size in GB
    encrypted             = true   # Encrypt the volume
    delete_on_termination = true
  }

  tags = {
    Name = "coder-${data.coder_workspace_owner.me.name}-${data.coder_workspace.me.name}"
    # Required if you are using our example policy, see template README
    Coder_Provisioned = "true"
  }
  lifecycle {
    ignore_changes = [ami]
  }
}

resource "aws_instance" "dev" {
  ami               = data.aws_ami.ubuntu.id
  availability_zone = "${data.coder_parameter.region.value}a"
  instance_type     = data.coder_parameter.instance_type.value
  subnet_id         = "subnet-xxxxxxxxxxxxxxxx"
  associate_public_ip_address = true

  user_data = data.cloudinit_config.user_data.rendered

    # Define the root volume storage
  root_block_device {
    volume_type           = "gp3"  # You can choose gp2, gp3, io1, etc.
    volume_size           = data.coder_parameter.disk_space.value     # Specify the volume size in GB
    encrypted             = true   # Encrypt the volume
    delete_on_termination = true
  }

  tags = {
    Name = "coder-${data.coder_workspace_owner.me.name}-${data.coder_workspace.me.name}"
    # Required if you are using our example policy, see template README
    Coder_Provisioned = "true"
  }
  lifecycle {
    ignore_changes = [ami]
  }
}

7 replies

CCoder.com

•Created by Chris2pi on 11/24/2024 in #help

Creating a JupyterLab instance

It might be simple, but I can't get a JupyterLab instance going. The goal is to spin up an AWS EC2 instance. I can do that with VSCode without issue. I add the module from here but get a few errors on launch - https://registry.coder.com/modules/jupyterlab Then I see this doco and it seems to be what the module is meant to do? https://coder.com/docs/admin/templates/extending-templates/web-ides#jupyterlab Can't get that working either. I suspect I am confusing the above doco and just using the module incorrectly and mixing it with the vscode out of the box template. Code below

resource "coder_agent" "main" {
  os   = "linux"
  arch = "amd64"
  dir  = "/home/coder"

  startup_script = <<-EOT
    set -e

    pip3 install jupyterlab

    # install and start JupyterLab
    ~/.local/bin/jupyter lab --ServerApp.base_url=/@${data.coder_workspace_owner.me.name}/${data.coder_workspace.me.name}/apps/jupyter --ServerApp.token='' --ip='*'
  EOT
}

module "jupyterlab" {
  source   = "registry.coder.com/modules/jupyterlab/coder"
  version  = "1.0.23"
  agent_id = coder_agent.main.id
}

locals {
  linux_user = "coder"
  user_data  = <<-EOT
  Content-Type: multipart/mixed; boundary="//"
  MIME-Version: 1.0

  --//
  Content-Type: text/cloud-config; charset="us-ascii"
  MIME-Version: 1.0
  Content-Transfer-Encoding: 7bit
  Content-Disposition: attachment; filename="cloud-config.txt"

  #cloud-config
  cloud_final_modules:
  - [scripts-user, always]
  hostname: ${lower(data.coder_workspace.me.name)}
  users:
  - name: ${local.linux_user}
    sudo: ALL=(ALL) NOPASSWD:ALL
    shell: /bin/bash

  --//
  Content-Type: text/x-shellscript; charset="us-ascii"
  MIME-Version: 1.0
  Content-Transfer-Encoding: 7bit
  Content-Disposition: attachment; filename="userdata.txt"

  #!/bin/bash
  sudo -u ${local.linux_user} sh -c '${try(coder_agent.main.init_script, "")}'
  --//--
  EOT
}

resource "coder_agent" "main" {
  os   = "linux"
  arch = "amd64"
  dir  = "/home/coder"

  startup_script = <<-EOT
    set -e

    pip3 install jupyterlab

    # install and start JupyterLab
    ~/.local/bin/jupyter lab --ServerApp.base_url=/@${data.coder_workspace_owner.me.name}/${data.coder_workspace.me.name}/apps/jupyter --ServerApp.token='' --ip='*'
  EOT
}

module "jupyterlab" {
  source   = "registry.coder.com/modules/jupyterlab/coder"
  version  = "1.0.23"
  agent_id = coder_agent.main.id
}

locals {
  linux_user = "coder"
  user_data  = <<-EOT
  Content-Type: multipart/mixed; boundary="//"
  MIME-Version: 1.0

  --//
  Content-Type: text/cloud-config; charset="us-ascii"
  MIME-Version: 1.0
  Content-Transfer-Encoding: 7bit
  Content-Disposition: attachment; filename="cloud-config.txt"

  #cloud-config
  cloud_final_modules:
  - [scripts-user, always]
  hostname: ${lower(data.coder_workspace.me.name)}
  users:
  - name: ${local.linux_user}
    sudo: ALL=(ALL) NOPASSWD:ALL
    shell: /bin/bash

  --//
  Content-Type: text/x-shellscript; charset="us-ascii"
  MIME-Version: 1.0
  Content-Transfer-Encoding: 7bit
  Content-Disposition: attachment; filename="userdata.txt"

  #!/bin/bash
  sudo -u ${local.linux_user} sh -c '${try(coder_agent.main.init_script, "")}'
  --//--
  EOT
}

Overall a end to end template would be nice. Happy to put one together once I understand it better.

7 replies

Gaming

Programming