I’m working through developing some production infrastructure for our container environment.
At the moment we have:
1x beefy physical compute node.
1x iSCSI san target over 1Gbps network.
I’m trying to decide between persistent storage on btrfs (which we have been successful with) and direct-lvm (which we haven’t used yet)
I’ve heard bits and pieces that a single direct-lvm block device can be between multiple docker nodes but are not finding any docs on it (which is a red flag)… is that true? lvm has clustered abilities baked in, and if docker is only using the persistent device on a single node at a time… it might work?
I like the ability to backup our btrfs persistent volume stores via rsync, I think we’ll have to backup direct-lvm volumes from within containers? (aka no direct host access?)
Obviously, we’ll be limited to mounting the storage on a single node if we go with a filesystem based backing like btrfs.
We’re too small of a shop to use kubernetes.