cwb-chef-repo

This Chef repo provides cookbooks to automatically install and configure Cloud WorkBench.

Requirements

Interested in your own Cloud WorkBench (CWB) installation?
These 10 steps will setup and configure CWB on AWS for you in less than 30 minutes (tested 2020-04-21) and costs ~1$ daily using two t3.small instances.

• Git

• Vagrant (2.2.4)

  • vagrant-omnibus (1.5.0) for auto-installation via Chef

  • vagrant-aws (0.7.2) for deployment in the Amazon EC2 Cloud (alternative providers are available)

  • Install Vagrant with the official installer or with Homebrew Cask via brew cask install vagrant

  • Install Vagrant plugins via

    vagrant plugin install vagrant-omnibus vagrant-aws;

• Amazon EC2 account. Alternative providers are available (see cwb-benchmarks#Providers). We have also deployed a CWB instance to OpenStack.

  • Both VMs (chef-server + cwb-server) must have a public IP address
  • Make sure you have created a private SSH key called cloud-benchmarking to log into cloud VMs and uploaded the corresponding public key to the cloud provider.
  • Ensure that incoming and outgoing traffic is allowed for ssh (22), http (80), and https (433). In Amazon EC2, you create a security group called cwb-web. If you do not explicitly specify a security group in your benchmark, make sure the default security group allows incoming ssh (22).

• ChefDK (3.9.0) for benchmark cookbook development.

  • Install via the official installer

Installation

NOTE: Checkout the Makefile which automates many of these steps if you are familiar with the configuration.

1. Checkout repository.

   git clone https://github.com/sealuzh/cwb-chef-repo;

2. Navigate into the appropriate install directory.

   cd install/aws          # Amazon EC2 Cloud (recommended)
   cd install/azure        # Microsoft Azure Cloud
   cd install/openstack    # OpenStack Cloud with public IP
   cd install/virtualbox   # Virtualbox (only for local development, unless you have public IPs)

3. Configure Vagrantfile and copy your private ssh key (for AWS) into cloud-benchmarking.pem.
   

   # For Amazon EC2
   AWS_ACCESS_KEY = ENV['AWS_ACCESS_KEY'] || 'my_aws_access_key'
   AWS_SECRET_KEY = ENV['AWS_SECRET_KEY'] || 'my_aws_secret_key'
   # SSH Key
   SSH_KEY_PATH = ENV['SSH_KEY_PATH'] || 'cloud-benchmarking.pem'
   SSH_KEY_NAME = ENV['SSH_KEY_NAME'] || 'cloud-benchmarking'

   • The private key will be copied into the cwb-server (/home/apps/.ssh/cloud-benchmarking.pem) for provisioning cloud VMs.
   • Find the aws config under config.vm.provider :aws (e.g., instance type).
   • Find the cwb-server config under chef.json. See providers for details how to configure other cloud providers.

4. Start automated installation and configuration.

   WARNING: This will acquire 2 VMs: one for the Chef Server and one for the CWB Server. Make sure you stop/terminate the VMs after usage in order to avoid unnecessary expenses.

   vagrant up

5. Once the Chef Server completed provisioning (around 4' on a t2.small instance) with
   INFO: Report handlers complete,
   update the public IP of chef-server (assigned by your provider)

   • Automatic query

     vagrant ssh chef-server --command 'wget -qO- http://ipecho.net/plain; echo' | tee chef_server_ip.env

   • Manual lookup (e.g. in your AWS Console) and save it to the file chef_server_ip.env

6. Setup the Chef Server

   1. Create the cwb-server admin user (replace chefadmin with a password of your choice)

      vagrant ssh chef-server --command 'sudo chef-server-ctl user-create cwb-server CWB Server [email protected] chefadmin' | tee cwb-server.pem

   2. Create the chef-validator organization

      vagrant ssh chef-server --command 'sudo chef-server-ctl org-create chef "CWB Chef" --association cwb-server' | tee chef-validator.pem

   3. Restrict file permissions with

      chmod 600 cwb-server.pem

7. Configure Chef knife and Berkshelf berks tools

   1. Within knife.rb, update CWB_CHEF_REPO, CWB_BENCHMARKS, and ENVIRONMENT.

   2. Symlink knife.rb to $HOME/.chef/knife.rb and config.json to $HOME/.berkshelf/config.json

      mkdir -p $HOME/.chef; ln -s "$(pwd -P)/knife.rb" $HOME/.chef/knife.rb;
      mkdir -p $HOME/.berkshelf; ln -s "$(pwd -P)/config.json" $HOME/.berkshelf/config.json;

8. Upload basic benchmark to the Chef Server

   cd $HOME/git;
   git clone https://github.com/sealuzh/cwb-benchmarks && cd cwb-benchmarks/cli-benchmark;
   berks install && berks upload;

9. Once the CWB Server completed provisioning (around 9' on a t2.small instance), reprovision to successfully complete the configuration (around 1').

   cd $HOME/git/cwb-chef-repo/install/aws/
   vagrant provision cwb-server

10. Browser to http://my_public_ip_of_cwb_server (http://33.33.33.20 for virtualbox) and login with the default password demo

    # Query public cwb-server IP
    vagrant ssh cwb-server --command 'wget -qO- http://ipecho.net/plain; echo'

Next steps:

1. Run a sample benchmark: https://github.com/sealuzh/cwb-benchmarks#execute-a-basic-cli-benchmark
2. Write your own benchmark: https://github.com/sealuzh/cwb-benchmarks

Workstation

This option automatically configures configures the browser-based IDE theia with the theia-ruby-extension for authoring CWB benchmark cookbooks. Notice that authentication is NOT supported for this optional component.

1. Configure the desired number of editor instances via NUM_WORKSTATIONS = 1 in the Vagrantfile
2. Setup everything via vagrant up
3. Access the web-editor via http://WORKSTATION_IP

Deployment

make deploy

Automates reprovisioning of the cwb-server for triggering deployment:

cd $HOME/git/cwb-chef-repo/install/aws/
vagrant provision cwb-server

Manage VMs

Acquire 2 VMs and install cwb-server and chef-server.

vagrant up

SSH into a VM (default: cwb-server)

vagrant ssh
vagrant ssh chef-server

Provision VMs

vagrant provision
vagrant provision cwb-server
vagrant provision chef-server

Sync folders (i.e., update cookbooks)

vagrant rsync
vagrant rsync cwb-server
vagrant rsync chef-server

Halt (i.e., stop) VMs

vagrant halt
vagrant halt cwb-server
vagrant halt chef-server

Destroy (i.e., terminate) VMs

vagrant destroy
vagrant destroy cwb-server
vagrant destroy chef-server

Refer to Vagrant CLI for further commands.

Reconfiguration on IP Address Change

make config_cwb

This make target automates the following steps:

1. Update chef-server IP

   • Automatic query

     vagrant ssh chef-server --command 'wget -qO- http://ipecho.net/plain; echo' | tee chef_server_ip.env

   • Manual lookup (e.g. in your AWS Console) and save it to the file chef_server_ip.env

2. Apply changes to cwb-server

   vagrant provision cwb-server

Manage Services

Precondition: SSH'ed into the cwb-server instance (if not using a make target locally)

Systemd

Foreman creates Systemd service templates upon deployment under /etc/systemd/system/. Background on How To Use Systemctl to Manage Systemd Services and Units

Targets

nginx.service

cloud-workbench.target
cloud-workbench-web.target
[email protected]
cloud-workbench-job.target
[email protected]
[email protected]
...

Status, Start, Stop, Restart

make cwb_status
make cwb_start
make cwb_stop
make cwb_restart

Automates the following cloud-workbench.target commands:

sudo systemctl status cloud-workbench.target
sudo systemctl start cloud-workbench.target
sudo systemctl stop cloud-workbench.target
sudo systemctl restart cloud-workbench.target
# Further examples
sudo systemctl stop cloud-workbench-job.target
sudo systemctl start [email protected]
sudo systemctl restart cloud-workbench-web.target

For further detail see: https://www.digitalocean.com/community/tutorials/how-to-use-systemctl-to-manage-systemd-services-and-units

View Logs

Precondition: SSH'ed into the target instance

Cloud WorkBench

make logs

Automates attaching to the cwb-server logs:

# Real-time (make logs)
journalctl -u cloud-workbench* -f
# Recent
journalctl -u cloud-workbench* -n 20

journalctl -u cloud-workbench*
journalctl -u cloud-workbench-web*
journalctl -u cloud-workbench-job*
journalctl -u [email protected]

tail -f /var/log/syslog

Benchmark Schedule Triggers

cat /var/www/cloud-workbench/shared/log/benchmark_schedule.log

Nginx

tail -f /var/log/nginx/cloud-workbench-access.log
tail -f /var/log/nginx/cloud-workbench-error.log

Installation directories (cwb-server)

# Rails app
cd /var/www/cloud-workbench/current
# Storage directory (where materialized Vagrantfiles are stored)
cd /var/www/cloud-workbench/shared/storage/production
# NGING proxy (sudo nginx -s reload)
cat /etc/nginx/sites-available/cloud-workbench
# PostgreSQL database (sudo su postgres)
ls /var/lib/postgresql/9.6/main
# Systemd service
ls -l /etc/systemd/system/cloud-workbench*

Rails Console

make cwb_console

Automates attaching to the cwb-server logs:

sudo su - apps
cd /var/www/cloud-workbench/current && RAILS_ENV=production bin/rails c

Backup

make backup

Automates the following backup process:

# Login into cwb-server
vagrant ssh cwb-server
# Stop server
sudo systemctl stop cloud-workbench.target
# Backup on cwb-server
sudo su - apps
cd /var/www/cloud-workbench/current
bin/rake data:backup
bin/rake data:list
# Start server
exit
sudo systemctl start cloud-workbench.target

# Download from cwb-server
exit
vagrant ssh-config cwb-server > ssh_config
scp -F ssh_config cwb-server:/var/www/cloud-workbench/shared/backups/*_cloud_workbench_production.* .

Restore

make restore CWB_BACKUP=backups/2019-01-07-18*cloud_workbench_production*

Automates the following restore process:

# Upload to cwb-server
vagrant ssh-config cwb-server > ssh_config
scp -F ssh_config *_cloud_workbench_production.* cwb-server:/home/ubuntu

# Login into cwb-server
vagrant ssh cwb-server
# Move files
sudo mv /home/ubuntu/*_cloud_workbench_production.* /var/www/cloud-workbench/shared/backups/ && sudo chown apps:apps /var/www/cloud-workbench/shared/backups/*
# Stop server
sudo systemctl stop cloud-workbench.target
# Restore from backup (purges current state!)
sudo su - apps
cd /var/www/cloud-workbench/current
bin/rake data:list # List backups
bin/rake data:restore[cloud_workbench_production] # File pattern argument or common date prefix (of .dump and .tar.gz)
# Start server
exit
sudo systemctl start cloud-workbench.target
# Check logs for errors
journalctl -u cloud-workbench* -f

PostgreSQL

Save login credentials via a password file:

echo "localhost:5432:cloud_workbench_production:postgres:rootcloud" > ~/.pgpass
chmod 0600 ~/.pgpass