Cary - ZFS - Caching - Replication

Hello,

I have three questions I've been meaning to ask.  Feel free to spread them out over multiple episodes if you want.

1) ZFS on ZFS.

This has come up several times recently, but I have a specific use case you might be able to comment on.  I'd like to have snapshots of my dropbox folder.  Since dropbox doesn't have a freebsd client, this means I need to use linux.  Using zfs on linux isn't a problem, but I'd also like to run this in a vm, ideally using bhyve, which means zfs on linux on a zvol (or file on zfs).  What's the best way to set that up?  NFS isn't an option since dropbox uses inotify, so it needs a local filesystem.  Should I just give up and either use dedicated hardware or keep using qemu/kvm on linux to run the vm?  Currently my dropbox folder is on btrfs for snapshots, and I'd like to move away from that.

2) Caching FreeBSD packages

I have a slow internet connection (3.5 mb/s) and would like to cache pkg files so that updates are faster.  There's a package that works with most linux distros called apt-cacher-ng that "just works" as a lazy mirror - the first time you download a package it reaches out to the internet and caches it, but subsequent times it uses the local copy.  It's also intelligent about expiring the cache for metadata files so updates still happen there even though the filename hasn't changed.  Is there any similar way to do this on FreeBSD?  For bonus points it would also work with OpenBSD packages and maybe even pkgsrc.

3) ZFS pull replication

I'm aware of zxfer and it's great, but I'd rather do pull replication so that backup host doesn't have to allow access from the clients.  Is there a script or relatively easy way to set up pull replication?

Thanks,
Cary