#Debian LAVA docker-container Designed to build and run a debian-based Docker container with LAVA on an Ubuntu 14.04 Host.

To build a new image locally

It may be desired to disable or add to the Dockerfile. This requires locally rebuilding the Docker Image. To build an image locally, execute the following from the directory you cloned the repo:

sudo docker build -t "lava_base:dockerfile" .

Where lava_base:dockerfile is the Docker image name and can be chosen at the time of build.

To run image

To run the container from a Ubuntu host terminal command line execute the following:

sudo docker run -it  -v /boot:/boot  -v /lib/modules:/lib/modules -v $PWD/fileshare:/opt/fileshare -v /dev/bus/usb:/dev/bus/usb  --device=/dev/ttyUSB0  -p 8000:80 -p 2022:22  -h de2384825135  --privileged=true arizidon/debian-lavaserver:testing

or

sudo docker run -it  -v /boot:/boot -v /lib/modules:/lib/modules -v $PWD/fileshare:/opt/fileshare -v /dev/bus/usb:/dev/bus/usb  --device=/dev/ttyUSB0  -p 8000:80 -p 2022:22  -h de2384825135  --privileged=true 4548d6b5a5b2

In the above command:

• -v /boot:/boot and -v /lib/modules:/lib/modules is neccessary due to libguestfs requireing a kernel and modules before it will run
• -v $PWD/fileshare:/opt/fileshare Includes a shared directory volume between the container and the host OS (./fileshare). Thus when invoking the docker run command, it should be done from a directory with ./fileshare in it.
• -v /dev/bus/usb:/dev/bus/usb opens up the USB ports so that fastboot can be connected to a physical target through USB.
• --device=/dev/ttyUSB0 assosiates the physical serial USB debug port so that a console can be accessed from the container command line using the pre-installed "screen" utility. Note that ttyUSB0 can change from host to host.

    screen /dev/ttyUSB0 115200        //  to exit screen type `<ctl-a> <ctl-d>`

• -p 8000:80 Opens up a network port to the host network so that the user can access the LAVA UI from a browser on another computer on the network <host network IP address>:8000
• -p 2022:22 opens up a ssh port 2022 for access from other hosts on the network to the container. The host:password is "root" and "password". Example: ssh root@<ip address of LAVA container host> -p 2022. This is mapped to 2022 for the use case where the Ubuntu host is already running sshd with the default port of 22 to prevent a conflict.
• -h de2384825135 -- with v2, LAVA needs to know the name of the worker machine.
• The final parameter is the image ID that the Container will be run. In the above example, 4548d6b5a5b2 is the resulting Image ID built in the previous step and will be different each time an image is built. The Image ID could also be a pre-built container such as arizidon/debian-lavaserver:testing used in the other example and comes from the prebuilt image that has been uploaded to docker hub . This pre-built image is the recommended image to install and test initially before building your own image.

A quick test

A script is included in the install to verify that this install was successful. It runs several QEMU tests. Once kicked off from the command line, progress of the tests can also be monitored from the LAVA Browser interface. To execute this command from the container command line, enter /submittestjob.sh.

#Alternative Configuration Options

Default Passwords

Default password for debian root is root:password. This is used for ssh as well.

Default admin account and password for the LAVA Server is admin:admin

NOTE: These should be changed in a deployed use-case

Create unique local network IP address

To this point, the procedure uses a specific port on the shared Host IP address on the local network (for example 192.168.0.1:8000). In some cases, it may be desirable for the LAVA Server to expose an entirely different IP address on the local network. To create a specific / secondary IP address to map to the LAVA Server on the host, the host can be configured as follows:

Reference: http://blog.codeaholics.org/2013/giving-dockerlxc-containers-a-routable-ip-address/

From Ubuntu Host OS

Define Virtual Interface

• use a macvlan type interface so that it could have its own MAC address and therefore receive an IP by DHCP if required.

sudo ip link add virtual0 link eth0 type macvlan mode bridge

• Use DHCP instead of selecting a static IP

sudo dhclient virtual0 &

Note: Must kill client when close container

sudo dhclient -d -r virtual0

• Bring it up

sudo ip link set virtual0 up

Set up inbound routing

sudo iptables -t nat -N BRIDGE-VIRTUAL0
sudo iptables -t nat -A PREROUTING -p all -d <container external IP> -j BRIDGE-VIRTUAL0

Where <container external IP> is the static OR DHCP defined IP address. Can find it using

sudo ifconfig
sudo docker inspect <container id>   // to find the IP
sudo iptables -t nat -A OUTPUT -p all -d <container external IP> -j BRIDGE-VIRTUAL0
sudo iptables -t nat -A BRIDGE-VIRTUAL0 -p all -j DNAT --to-destination <container internal IP>

Set up outbound Routing

sudo iptables -t nat -I POSTROUTING -p all -s <container internal IP> -j SNAT --to-source <container external IP>

Can now connect to LAVA through a browser using the from the local network

Configuration Control

Base Image Version Control

One very important step in using this LAVA server installation is backing up the Docker Image once it has been customized for a specific deployment. Once a Docker Base Image has been prepared for deployment (potentially across multiple servers), it is recommended that the image be placed in a repository on http:/hub.docker.com OR if the above does not comply with company security policies, then it will be required to create a local docker repository within the company infrastructure and back up to this repository.

Deployed Container Backups

Docker makes it very simple for the user to back up a running container that has been deployed and is configured. This is quite convenient if a host were to fail. The backed up image can effectively be restored and all configuration / test options will be present except those changed from the last backup. See thedocker export command for more details.

If for any reason a container is stopped using the exit command from the container command line, the state at exit can be saved from within the Ubuntu host using the docker commit command. Command format may look similar to the following:

sudo docker commit <containerID>  <newimage:tag>

See docker commit in the Docker documentation for more details.

Known Issues / Warnings

• Note that when the container is run, there is a privilege = true option in the command line. This is because LAVA is required privilege access to the OS to be able to mount loopbacks in order to function. The user should be aware that the whole container is running in privileged mode due to this.
• An intermittent issue has been seen. When running the /submittingtestjob.sh script which effectively runs an integrity test set of jobs against several QEMU installed targets, on occation an error occurs that looks similar below. It occurs when running the /submittestjob.sh script. If this occurs, it's an indicator that Apache2 has died. The LAVA Server Web UI will also likely be unresponsive. If this occurs, run service apache2 reload from within the running container and it should be resolved.

xmlrpclib.ProtocolError: ....500 Internal Server Error

• As noted in the previous bullet, the Dockerfile is currently configured to automotically install 3 different QEMU targets for testing. If this is not desired, the user should modify these configuration options from within the Dockerfile and rebuild to remove them from the resulting image.

Docker tags in this project

• arizidon/debian-lavaserver:testng #Initial test build
• akbennett/lava:debian-sid # Base Debian sid image