seaweedfs

1. add back recursively delete a folder
2. speed up leveldb bolddb loading
3. fix problem with master submit error "multipart handled by ParseMultipartForm".
4. default to v3 volume format
5. add master client to listen for volume location updates
6. add 128 bit needle id. see https://github.com/chrislusf/seaweedfs/wiki/Optimization#customize-needle-id-to-128-bit
7. add filer notification on change events
8. assert master server count is odd
9. add redis cluster store for filer
10. filer UI add upload and drop & drop
11. add "weed scaffold" command to generate configuration files

Fix possible nil bug merged from a pull request.

Last release for version2 volume storage. Switching to version3 volume storage in next release.

1. faster loading volumes with boltdb or leveldb needle map
2. heartbeat to detect disconnected volume servers
3. filer add dir list size limit
4. mount a filer folder to a local folder
5. "weed mount" now supports changing attributes, e.g., chown, chmod
6. add "weed s3" to support Amazon S3 API, with basic PUT/GET/HEAD/DELETE bucket, PUT/GET/HEAD/DELETE objects

fixed an important bug: during compaction, there could be some deletion. In old version, there would be some error and the compaction always fails. This version handles the deletion well.

accumulated changes since last 12 months.

The next release will default to a new filer implementation with more POSIX-like features. Current filer users may want to remember this version in case the upgrade is not so smooth.

1. Add "weed volume -port.public=xxx" option for read only access.
2. Change from "publicIp" to free-text "publicUrl".
3. Change master if master leader changes.
4. Correctly handle .tar.gz file uploads, avoiding auto unzip.
5. Avoid wrong way to handle write failure for replicated writes, which could lead to volume .dat file format error.
6. Support in-memory/leveldb/boltdb mode tuning for memory/performance balance.
7. Started basic admin ui.
8. Add option for filer to disable directory listing.