To do:

 - graceful handling of nearly full snapshot stores
 - snapshot and replicate commands should wait on ok/err message?
 - track filesystem information (type, mount options)
 - freeze filesytem before taking snapshots
 - use sudo to not require root keys
 - better success/failure messages
 - many FIXME TODO's in code, many input validation
 - recover from interrupted replication

Done:

 - skip this replication cycle if no chunks have been updated
 - cleanup unreferenced snapshots on replication failure and remove send file
 - get rid of Pseudo-terminal will not be allocated because stdin is not a terminal. message
   (as well as stdin error, 'echo cmd | ssh -T host sh')
 - forget volume needs to clean up sources and targets
 - log consistently and with correct pid

 - snapshot rotation
 - init script
 - snapshot, replication daemons
 - admin database editing commands
 - Hook up master snapshot schedule to cron
 - set up development tree
 - install/uninstall/develop scripts
 - create/delete ddsnap sockets
 - start ddsnap daemons
 - create/delete devmapper device on snapshot create/delete
 - mount/unmount device on snapshot create/delete
 - add use counts

 - Remote operations
    - ssh to start downstream listen
       - confirm downstream holds expected snapshot
    - ssh to set new downstream snapshot
    - downstream node kicks upstream on zumastor start via ssh

 - make "receive" commands log locally