Update Terraform to automate host and server setup

[emilyllim]

Existing Terraform configuration should be updated to automatically install everything the instance needs to be set up as a runner.

Resources

• Terraform Documentation: user_data
• Adding self-hosted runners

[emilyllim]

Updates added to existing pull request #7

[emilyllim]

Updating the Terraform configuration

Before adding the setup commands, I tested the use of the user_data attribute by creating a test script that just echoes some text to a new file. Then I applied the updated Terraform configuration:

• terraform apply

Then I SSH'ed into the instance. At first I was confused as to why I couldn't find the new file, but it was created at the root directory level, not the user (ubuntu) level.

• cd /
• ls - the file should be viewable at this level

[emilyllim]

Add a self-hosted runner

First, I manually followed the instructions in the GitHub documentation to add a self-hosted runner for the cinder_bench_runner repository:

• Settings -> Actions -> Runners -> New self-hosted runner
• I selected the operating system and architecture of my EC2 instance (Linux and x64, respectively)
• The commands intended for OS and architecture chosen are provided by GitHub
• Ran into an error when running the command to create the runner:
  • ./bin/libcoreclr.so: /usr/lib/x86_64-linux-gnu/libstdc++.so.6: version 'CXXABI_1.3.7' not found (required by ./bin/libcoreclr.so) Libicu's dependencies is missing for Dotnet Core 6.0
• Output tells me to run sudo ./bin/installdependencies.sh
• But I ran into more errors:
  • E: Some index files failed to download. They have been ignored, or old ones used instead. 'apt-get' failed with exit code '0' Can't install dotnet core dependencies.
• I tried sudo apt-get update, but continued running into several Failed to fetch errors

[emilyllim]

Troubleshooting "Failed to fetch" errors

• I tried to set up a runner on another machine and it worked, so this was not an issue on my own network.
• I tested the network of the EC2 instance by using ping www.google.com:
  • I can see packets arriving in the output, so there isn't an issue with the internet connectivity.
• Next thing I checked was the /etc/apt/sources.list file.
  • It should have package repositories that match the Ubuntu version I am using (Precise Pangolin).
  • All of them matched, but I realized that the current AMI I am using is very old, so I decided to switch to a newer one.
• I changed the AMI to one that is running 22.04.3 (Jammy Jellyfish).
• I tried running sudo apt-get update and now there are no fetching errors!
• Then I manually ran the commands to set up the runner again.

Check that the runner has been added successfully

• In the repository, go to Settings -> Actions -> Runners. The runner should be present as "Idle", or "Offline" (if it hasn't been run yet).

Add commands to set up runner in a script

• I copied the commands I ran manually to the script that will be run by Terraform.
• There was one change needed to automate this process. For the runner creation command, I added arguments for the runner group, name, labels, and work:
  • --runnergroup "Default" --name "benchrunner2" --labels "self-hosted,Linux,X64" --work "_work"
  • Therefore, the command line won't ask for these values manually.
• Provisioning the resources through Terraform runs these commands and sets up the runner as in the previous steps.

(Optional) removing the runner

• In the repository, go to Settings -> Actions -> Runners -> Click the three dots and select "Remove" and follow the instructions.

[emilyllim]

Next step

• Before adding these changes to the repository, I need to secure the token as a secret and update the script to use this secret instead of hardcoding it.