richardsharpe / amazon-s3-vfs Goto Github PK

vfs_amazon_s3.

This is a work in progress. It is an experiment to see how well we can 
integrate Amazon's S3 with Samba via a VFS.

The idea is to allow a Samba share to be a place that you could use to drop
files into and then have them saved up on Amazon's S3 service as a backup for
you.

You need libxml2-devel and libcurl-devel.

To build (against the master branch, or a 3.6.x branch with the FSCTL patches):

1. ./autogen.sh
2. ./configure --with-samba-source=/path/to/samba/source3/
3. make

You need to add the following fields to the share you want to have push files
up to Amazon's S3 service:

[the-share]
	path = /some/path
	vfs objects = vfs_amazon_s3
	vfs_amazon_s3:access key = your-access-key
	vfs_amazon_s3:secret key = your-secret-key
	vfs_amazon_s3:bucket = the-name-of-the-bucket
	vfs_amazon_s3:reloc dir = /path/to/some/dir/not/under/the/share

Yes, I know, the secret key needs to be dealt with in a better fashion.

Progress.

The latest, as of 11-Feb-2012, is that it now uploads files. However, it only
uses PUTs to do that, and has a problem with the delimiter. It really should
use multi-part uploads for large files. It should probably also allow SDs to
be uploaded, perhaps with some simple encoding/serializing of the SD and the
data and any alternate data streams.

TODO:

1. Add tests to configure.in for libxml2-devel and libcurl-devel.

2. Fix bugs in upload and add support for multi-part uploads.

3. Add support for getting previous versions via FSCTLs.

4. Support opening and retrieving files up on S3.

5. Add support for uploading any SDs stored in XATTRs (or even permissions) 
   and perhaps alternate data streams.