v3.2.0 released
New features
- Added snappy compression support for Raft entires and snapshots.
- Added experimental witness support.
- Added new API to allow LogDB compaction to be manually triggered.
- Added event listener support to allow users to be notified for certain Raft events.
- Added system event listener support to allow users to be notified for certain system events.
- Added Raft related metrics to exported.
- Added rate limit support to control the maximum bandwidth used for snapshot streaming.
- Updated the C++ binding to cover all v3.1 features. Thanks JasonYuchen for working on that.
- Added a virtual filesystem layer to support more filesystem related tests.
- Added experimental Windows and FreeBSD support.
Improvements
- Removed the restriction on max proposal payload size.
- Re-enabled the range delete support in LogDB.
- Better handling of concurrent snapshot streams.
- Extensive testing have been done on a high performance native Go KV store called Pebble.
- TolerateCorruptedTailRecords is now the default WAL recovery mode in the RocksDB based LogDB.
Breaking changes
There is no breaking change for regular users. However,
- If you have built customized transport module implementing the raftio.IRaftRPC interface, there is minor change to the config.RaftRPCFactoryFunc type. See github.com/lni/dragoboat/config/config.go for details.
- The default transport module has been updated, it is no longer compatible with earlier versions of dragonboat.
- The default LogDB data format is no longer backward compactible with v3.1 or earlier.
Other changes
- LevelDB support has been marked as depreciated. It will be removed from dragonboat in the next major release.