I was thinking of setting up a 3 node LXD cluster with CEPH storage backend.
But I have tried googling if automatic live migration will happen in case a node or OSD goes down.
I did find one forum post where a poster was told that it’s beyond the scope of the LXD project.
Is this true?
If that is the case is it possible to script? Because I know live migration is supported I just haven’t found any documentation on automatic migration in case of node/storage failure.
The post you found is correct: that’s beyond the current scope of LXD clustering.
You can script that behavior by monitoring your nodes and triggering the appropriate measure when you detect that one of them is down.
Is the reasoning behind this that LXD is not a orchestrator?
Should I be looking on some other solution for my use case?
I want a container cluster with HA.
Yes, the reasoning is that LXD clustering is not meant to get into orchestration territory, at least for now. You should think of LXD clustering as a vSphere replacement, where instead of VMs you use system containers, which get you more efficient resource utilization.