Network and storage cloning in MAAS
After 3 iterations of UX and design phases, our work on network and storage cloning has come to the QA phase of our development cycle. In the up-coming release, you will be able to clone multiple destination machines.
Cloning works by selecting multiple machines from the machine listing page. These are going to be your cloning destination machines. Then click
clone from to select the source machine. There is a small caveat that your destination machines can only be in one of these 2 states; failed testing or ready. However, your source machine can be in any state.
You can search for your source machine or select a source machine from the list. This machine list is ordered by last modified machines and will only include machines that are not selected as the destination.
Once you select either the network, storage or both, MAAS will clone your source machine into your destination machines.
If there is a mismatch between boot methods, network interface, block device, or if the disk size of the source machine is smaller than the destination machine, the output will throw an error for those machines. A machine may have multiple errors after cloning, therefore, we grouped all machines with the same error together. When ‘show’ is click, it will filter the machines that are affected by the error.
LXD Cluster support: UI exploration in MAAS
In this iteration, we have been working on supporting the LXD cluster attributes in MAAS. In LXD, the concept of project is a given permission for a user to access resources from a LXD server or a group of LXD servers (LXD Cluster). Our goal is to allow people to register their projects in MAAS. If that project belongs to a LXD cluster, we will show all the LXD nodes that you have access to in that project. If it belongs to only 1 server, it should show a single LXD server that it has access to.
With this goal in mind, we have explored a few options in the UI to efficiently show that a project belongs to a cluster.
The first version we have explored is to revamp our current KVM listing table to show all the VM hosts that a project has access to.
We have also looked at an alternative option to make it look cleaner in case there are more than 1 VM host.
However, we’ve realised that the current way of showing this information is not very scalable. This is because a cluster can have up to 50 VM hosts, which is an edge case that can happen. This means that a user will be tired scrolling through pages to explore their options.
We’ve also thought about an alternative design that will allow us to show all the projects, yet list all the VM hosts in a way that is easier to digest.
This alternative view requires you to select a project by default and you will have a listing view of all the VM hosts that you have access to with the aggregate resource information of a cluster.
These are not our final design. We are still looking at different alternatives that would be more efficient to show. Any feedback regarding our design or how you view LXD cluster from a MAAS perspective is more than welcome at this stage. Let us know what you think.