Open johananl opened 4 years ago
Hm, I think this should have new upstream issue opened, as it heavily involves bootkube.
I would also suggest to specify, what should happen, when bootkube is re-executed, as I guess the expectations may vary.
In my case the problem is that /home/core/assets
was there but the target already existed and had to be cleaned up manually first. I think this mv
logic on our side can be made more robust (maybe this interaction of file
provisioner and the remove-exec
could be simplified, too).
In my case the problem is that /home/core/assets was there but the target already existed and had to be cleaned up manually first. I think this mv logic on our side can be made more robust (maybe this interaction of file provisioner and the remove-exec could be simplified, too).
Bootkube will fail when resources already exist, which is a major problem to solve, if we want to solve it. Moving files around is a nit.
Right now if a cluster bootstrap process fails due to Bootkube being unable to complete for any reason, rerunning
lokoctl cluster apply
results in an error:It could be much nicer to make the Bootkube part of our stack idempotent as well. This will likely require changes to Bootkube itself. Relevant upstream issue: https://github.com/kubernetes-sigs/bootkube/issues/700