rear-automated-testing

Relax-and-Recover (ReaR) Automated Testing is using Vagrant and standard GNU/Linux boxes to deploy one server and one client virtual machine (VM), and besides a recover VM is also created, but not provisioned. The main goal of this project is to foresee in a simple way to test the latest snapshot release of ReaR without too many manual interactions by doing a full backup with an automated restore into the recover VM.

In short we start the client and server VM via vagrant, and do a provisioning if required so that the server VM is capable of being a NFS server, PXE server and TFTP server. Then, we install the latest ReaR snapshot in the client VM and run a full backup using the NETFS backup method (with the help of tar). The tar archive is stored on the server VM and in the same time the PXE environment is configured on the server VM as well (when using KVM/libvirt), or a PXE environment is configured on the host system when using virtualbox.. When the ReaR backup is completed we halt the client VM and start the recover VM and do a full restore of the client content. Once the restore is completed the recover VM reboot automatically.

Clone this Git repository

$ git clone https://github.com/gdha/rear-automated-testing.git

Execute the automated ReaR Recovery test

Select the GNU/Linux OS to test by going into the proper directory. However, initially we only have centos7, but more will follow over time (and do not forget it goes much faster with sponsoring).

$ cd rear-automated-testing
$ sudo ./rear-automated-test.sh -h

+--------------------------------------------------+
|    Relax-and-Recover Automated Testing script    |
|             version 1.0                          |
+--------------------------------------------------+

Author: Gratien D'haese
Copyright: GPL v3


Usage: rear-automated-test.sh [-d distro] [-b <boot method>] [-s <server IP>] [-p provider] [-c rear-config-file.conf] [-t test] -vh
        -d: The distribution to use for this automated test (default: centos7)
        -b: The boot method to use by our automated test (default: PXE)
        -s: The <boot server> IP address (default: 192.168.33.15)
        -p: The vagrant <provider> to use (default: virtualbox)
        -c: The ReaR config file we want to use with this test (default: PXE-booting-example-with-URL-style.conf)
        -t: The ReaR validation test directory (see tests directory; no default)
        -h: This help message.
        -v: Revision number of this script.

Comments:
--------
<distro>: select the distribution you want to use for these testings
<boot method>: select the rescue image boot method (default PXE) - supported are PXE and ISO
<boot server>: is the server where the PXE or ISO images resides on (could be the hypervisor or host system)
<provider>: as we use vagrant we need to select the provider to use (virtualbox, libvirt)
<rear-config-file.conf>: is the ReaR config file we would like to use to drive the test scenario with (optional with PXE)
<test-dir>: under the tests/ directory there are sub-directories with the beakerlib tests (donated by RedHat).
       When -t option is used then we will not execute an automated recover test (at least no yet)

Using KVM/libvirt

To use this tool on a KVM/libvirt Linux system and as CentOS7 and PXE are the default (so far) we only require one parameter and that is -p libvirt as virtualbox was choosen as the default provider.

Using VirtualBox

To use this tool on a VirtualBox Linux system and as CentOS7 and PXE are the default (so far) we do not need additional parameters as it will try to set-up and FTPboot area on the host system itself. However, it is imported that the hosts system is a NFS server as the client VM will try to mount the TFTboot area. Why do we need the host system? That is because VirtualBox uses a NAT network to PXE boot from and that is always pointing the host system (a pity we cannot use the server VM).