Skip to content

README.md

Latest commit

 

History

History
70 lines (45 loc) · 3.27 KB

README.md

File metadata and controls

70 lines (45 loc) · 3.27 KB

Usage Guide

What does this do?

This folder contains ansible playbooks to install dependencies and deploy Xinda. We use it to configure test nodes on CloudLab.

Specifically, below are the things that this folder do:

  1. Installation
    • docker and docker-compose
    • blockade: installed in a conda environment with python3.6. The env is automatically activated when you login to the node.
    • Thrift and the correct toolchain to build it.
    • Charybdefs and the correct toolchain to build it: the built directory is copied to ~/workdir/xinda-software/charybdefs.
  2. Setup environment variables in ~/.bashrc.
  3. Clone Xinda and Xinda-software repos to ~/workdir.

Note

  1. Most of the installed programs' binaries are either installed in or linked to /usr/local/bin. So even if you don't activate the conda environment, you can still use the blockade command.
  2. Some of the environment variables might not be available in a non-interactive shell. Do mind that when you run the experiments (try to always use interactive shells).
  3. Important All playbooks are supposed to be run on a c220g2 node (available in the wisconsin cluster). The c220g2 node has a 480GB SATA SSD mounted at /var/lib/docker. The workdir is mounted on /dev/sdb, a 1.1TB HDD.

How to use it?

Step0: Create nodes on CloudLab

Create nodes on CloudLab with a Ubuntu 18.04 profile. A detailed instruction on how to use CloudLab can be found here.

Step1: Setup ssh key forwarding

Execute the following command to start the ssh-agent on your local machine.

eval `ssh-agent -s`

Then you should add your private key that has access to the xinda and xinda-software repos to the agent.

ssh-add <path-to-private-key>

Step2: Install ansible

We use Python==3.9.20 and pip==24.2 to install ansible in our local environment.

pip3 install ansible
ansible-playbook -h

Step2: Create your ansible_host file

The ansible_host file is used to specify the hosts that you want to run the ansible-playbook on. The format of the file is as follows:

c220g2-011110.wisc.cloudlab.us ansible_connection=ssh ansible_user=YOUR_USERNAME ansible_port=22

You can add as many hosts as you want to the file (i.e., a cluster of test nodes). The ansible_user is the username that you use to login to the host (your CloudLab username). The ansible_port is the port that you use to ssh to the host. The ansible_connection is the connection type that you use to connect to the host. In our case, we use ssh.

Step3: One-click setup using ansible-playbook

ansible-playbook -i ansible_host configure.yml

The configure.yml file is the entry playbook that runs a collection of playbooks in this directory. You can also run individual playbooks by specifying the corresponding file name.

The setup process will take ~30 minutes to complete (depending on your Internet connection). The script will install all necessary dependencies for Xinda to deploy a distributed system, run benchmarks, inject slow faults, collect runtime logs and stats, and analyze the results. Once the setup is done, you can ssh into the remote machine and start using Xinda (by default, the code is located at ~/workdir/xinda).