This repository contains the scripts and documentation to easily deploy CERNBox client-testing configurations with smashbox framework. Analysis and visualization of test results can be monitor with kibana using the kibana-plugin in smashbox/python/monitoring/kibana_monitoring.py.
We deploy smashbox with the following purposes:
- Regression-testing
- Behaviour comparison of the sync client with different configurations: platforms, cernbox client versions and endpoints
- Test the sync client installation on different platforms (e.g., Windows, CentOS, MacOSX)
Currently, it is possible to deploy smashbox in your current machine, a cluster of VMs in OpenStack or within a set of containers using Docker. Finally, this document also describes how to visualize and analyse the test results with kibana (instructions)
This repository is organised in the following way:
smashbox-deployment
├── documentation/ : procedures to manually set up a machine for continuos testing and monitoring
├── docker/ : scripts, docker files and configuration used to automatically deploy and continuosly run smashbox tests in a set of containers
│ └── Dockerfiles/ : dockerfiles used to build different images for each platform
│ └── setup.d : these are the main scripts used to deploy the specified architecture
├── kibana/ : this folder contains json files that stores kibana dashboards configurations
├── setup/ : this folder contains setup scripts and tools to automatically setup the machine
└── auto-smashbox.conf : default configuration file for smashbox
└── cbox_vers.json : json table which contains all available cernbox client versions for each platform
└── cleanup.py : script to clean up space in Windows
└── cleanup.sh : script to clean up space in Linux/Mac
└── get_vers.py : script to extract correct cernbox version form json table
└── linux_distr.py : script to extract OS platform version
└── setup.sh : setup script for Linux-CC7
└── ubuntu-setup.sh : setup script for Linux-Ubuntu
└── update_repo.py : script to update smashbox repositor in Windows
└── win-setup.py : setup script for Windows
└── README : this file
** cbox_vers.json is a table with 3 columns. 1st column is version for Linux, 2nd for MacOS and 3rd for Windows.
- Deploy and set up a testing cluster of VMs (Openstack)
- Deploy and set up a testing cluster of containers (Docker)
- Monitoring and Analysis with kibana
If you want to set up a machine for continuos testing and monitoring with smashbox, you can execute the corresponding setup script depending your OS. This script is developed to automatically and dinamically install the OwnCloud client, configure smashbox and install the cron job. The steps to use these scripts are the following:
** After creation of a new Windows VM you have to contact CDA to remove the machine form their automatic cernbox client update campaign.
Python 2.7 is required. You can download it here: https://www.python.org/downloads/windows/
You can download and install git here: https://git-scm.com/download/win
In Windows search tab search for 'Git Bash', right click and open as administrator
In git bash run: git clone https://github.com/cernbox/smashbox-deployment.git
In win-setup.py, 2 variables run_time and cleanup_time indicate what time should the smashbox and cleaup processes run. You can modify them to the time you want.
In Windows search tab search for 'Command Prompt', right click and open as administrator
Locate where you cloned the smashbox-deployment folder, from cmd cd in the smashbox-deployment/setup/ folder
From cmd run the following command:
C:\Python27\python.exe win-setup.py -v VERSION -u USERNAME -p PASSWORD -k KIBANA_ACTIVITY
Python 2.7 is required. If your machine is an openstack VM then python 2.7 is already installed. Else you need to download and install python 2.7.
In terminal run the following command: sudo yum install git
In terminal run the following command: git clone https://github.com/cernbox/smashbox-deployment.git
In terminal: cd smashbox-deployment/setup/
In terminal run the following command: ./setup.sh -v VERSION -u USERNAME -p PASSWORD -k KIBANA_ACTIVITY
** VERSION is the cernbox client version you want, USERNAME/PASSWORD are your credentials for cernbox and KIBANA_ACTIVITY is variable so 'Kibana-Monitoring' can identify the data that are sent. It is described bellow in monitoring section.
If you want to set up the cluster with containers. You should run the following commands:
docker build -t debian-smashbox
docker run -it -e SMASHBOX_OC_ACCOUNT_NAME="XXXX" -e SMASHBOX_OC_ACCOUNT_PASSWORD="YYYYYY" -e SMASHBOX_OC_SERVER="cernbox.cern.ch" debian-smashbox:latest bash
The docker build should make reference to the dockerfile with the image desired to set up. Then, it is also neccesary to define the following environment variables:
SMASHBOX_OS
SMASHBOX_CLIENT_VERSION
SMASHBOX_OC_SERVER
SMASHBOX_OC_ACCOUNT_NAME
SMASHBOX_OC_ACCOUNT_PASSWORD
SMASHBOX_TESTDIR
SMASHBOX_SSL_ENABLED
The goal of this section is to provide a convenient monitoring tool for the deployed smashbox testing architecture. For this purpose it has been choosen kibana for visualization and elasticsearch to store tests results.
In order to visualize the test results in kibana; first it is needed to provide the following parameters in smashbox/etc/smashbox.conf
- kibana_monitoring_host. This is the host machine where you have running kibana. For example: kibana_monitoring_host = "http://monit-metrics"
- kibana_monitoring_port. This is the port to communicate with ELK. For example: kibana_monitoring_port = "10012"
- kibana_activity. This is an additional parameter to be able to identify the data that you are sending to ELK. For example: kibana_activity = "smashbox-regression"
The kibana web interface is accessible in https://monit-kibana.cern.ch.
If you don't have yet the dashboard configured. You can download the json file kibana\cernbox-smashbox.json; then you need to go to the tab "management" in kibana and import this json file as a saved of object.
The dashboard has been designed to monitor the failed tests running smashbox with different OwnCloud client versions and different platforms. For regression testing, tests are executed periodically according to the csv file provided in the deployment and the schedule specified there.
** Note: This section is based on the current deployed ELK architecture at CERN. In order to easily deploy an ELK architecture there is a document describing the procedure kibana/elk-docker.pdf.