⚠️ This repository is deprecated! A much better implementation is officially maintained as a part of LAVA at git.lavasoftware.org.
This repository attempts to provide a reference implementation of deploying LAVA using its officially distributed docker containers.
This lava-docker-compose repository is designed to be the simplest possible way to for anyone to run LAVA locally, and provides a qemu worker by default.
To see a more elaborate example with actual boards, see lava.therub.org.
Install the following.
No configuration should be necessary when running a simple qemu worker.
make: Deploy a pgsql container, lava server container, lava dispatcher
container, and image host container. A user (username admin, password admin)
will automatically be deployed, as well as a qemu device-type and a qemu-01
device. Its health check should run automatically.
make clean: Permanently delete the containers and the pgsql volume.
Once up, go to your http://localhost (port 80) and log in with admin:admin. You should see qemu-01's health-check running and it should finish successfully.
docker-compose down- Back up pgsql from its docker volume
sudo tar cvzf lava-server-pgdata-$(date +%Y%m%d).tgz /var/lib/docker/volumes/lava-server-pgdata
- Change e.g.
lavasoftware/lava-server:2019.04tolavasoftware/lava-server:2019.05andlavasoftware/lava-dispatcher:2019.04tolavasoftware/lava-dispatcher:2019.05in docker-compose.yml. docker-compose up
LAVA 2019.07 introduced a
change that
requires some overlayed files to be owned by the 'lavaserver' user. An
entrypoint.d file called chown-files.sh has been introduced to chown the
mounted files. This has the downside of changing the ownership of the files
checked into git, which is modestly inconvenient.
This upgrade changed the uid and gid of the lava user in the container to 200:200. After upgrading, run the following command to chown /var/lib/lava-server accordingly:
$ docker-compose exec server chown -R lavaserver:lavaserver /var/lib/lava-server/
The design goal of this repository is to demonstrate how to use the official LAVA docker containers in a native way. Ideally, this means without having to rebuild or modify the containers in any way. In the events that the containers or their entrypoints do need modifications, patches should be pushed upstream to provide interfaces for the given functionality. As the official containers mature, this repository should become more simple.
There are 4 containers defined in docker-compose.yml:
This is an official postgres container, and runs using a docker volume. Using an official postgres container is a lot easier than using the postgres that comes with the lava-server container. This set-up makes data persistant by default, and makes backups easy (if desired).
Note that ./server-overlay/etc/lava-server/instance.conf provides connection details to this database for lava-server.
This is a squid container that serves as an http proxy to the LAVA dispatcher. Its purpose is to cache downloads to a docker volume, to improve performance and prevent duplicate downloads. This is enabled in ./server-overlay/etc/lava-server/env.yaml
This is lava-server (lava master). In order to provision a qemu device automatically, a script at ./server-overlay/root/entrypoint.d/provision.sh is run at boot time to add a superuser (admin/admin) and a qemu device and qemu worker.
Several other files are mounted into the container. settings.conf is provided, as well as device and health-check directories.
The lava dispatcher is run using the official container directly. However, to use an actual board container modifications would have to be made in a similar way as they were made to lava-server. See lava.therub.org for an example of an actual lab with multiple boards.