Snappy
Snappy is a simple tool for making directory snapshot. It uses a content-addressable filesystem inspiered by git, but is much simpler and provides no VCS functionality.
Use cases
Snappy is intended as a light weight tool to create snapshots. My personal use is for use with HPC workload managers like SLURM where I need to copy files to a separate file system for each job. With snappy I can take a snapshot when the job is created and be safe that the files won't change by the time the job runs.
Usage
To start, in the directory you want snapshot, run:
$ snappy init
This will create .snappy.
Snapshots
To create a snapshot:
$ snappy snapshot
This will return a sha256 hash that is the identifier of the snapshot. All files at the time the snapshot was created will be stored in .snappy/storage.
List all snapshots by running:
$ snappy ls
To delete a snapshot, run:
$ snappy rm <snapshot-hash>
Note that this will not remove the objects in storage, but only remove the directory tree described by the snapshot. To remove unrefrenced objects, run:
$ snappy clean
To view changes since the last snapshot:
$ snappy diff
Ignoring files
Snappy can ignore files just like git. Patterns can be added to .snappy/ignore. The pattern matching works like regular UNIX wildcard expansion.
Loading snapshots
Snapshots can be loaded by either copying out all the files, or by symlinking to the objects.
To copy files, run:
$ snappy checkout <snapshot-hash> [dst]
If dst is ommitted, the snapshot will be checked out in the working directory.
To link a snapshot, run:
$ snappy link <snapshot-hash> <dst>
This will create symlinks to all the files in the snapshot in the directory dst. This is good if you only need read-only access to the files.