Today the LAVA team was discussing how we could make LAVA aware of known-good images for each device type. The use case is: when learning LAVA, engineers will often not know exactly which images to use, and they usually don't care much as they want first to learn how to use LAVA itself.
To solve this problem, we came up with the following proposal.
The LAVA admin should be able to configure the URL of a mini-database of know good images for each of its device types. The proposed format is a very simple YAML hash like this:
----------------8<----------------8<----------------8<----------------- kvm: action: deploy_linaro_image parameters: image: "http://url.to/kvm.img" kvm-arm: action: deploy_linaro_kernel parameters: kernel: "http://url.to/zImage" rootfs: "http://url.to/rootfs" dtb: "http://url.to/dtb" arndale: action: deploy_linaro_image parameters: image: "http://url.to/arndale.img" ----------------8<----------------8<----------------8<-----------------
LAVA will just poll this database from time to time, but won't care how it is kept up to date. LAVA admins can make the generation of this database part of their automation.
In Linaro, we should probably generate this data from releases.linaro.org. Fathi, can we generate this as part of the release process? If not, can we at least have a MANIFEST of the full contents of each release, so that we can generate that database on our side? Assuming the directory strutcure of releases is stable, the output of `find .` inside the release directory should be good enough.