[06:20:59] <moritzm>	 the blocking is only an issue if one of those non-redundant VMs is down (like the etcd nodes), it's not an issue for the properly redundant other VMs. using the plain storage in production is really more of a hack than using Ganeti as designed
[06:23:16] <moritzm>	 wikikube already moved to running etc on the (baremetal) control plane nodes, for DSE the control plane servers are also VMs, but you could move the etcd for dse to the an-conf servers, which are on baremetal, already row-redundant and basically idle