v0.4.5-beta.1

Enabling ZSTD compression for stress testing

v0.4.4-RC

Enabling ARM builds

v0.4.3

• Timer efficiency improvements
• Timer based change log cleanup
• Improvements to documentation
• Removing redundant values from replication event
• Improving replicator to only create streams if they are missing, this
  will let admins operations persist on jetstreams in case of restart

v0.4.1

What's Changed

• Moving to NATS instead of dragonboat due to it's inefficiencies. by @maxpert in #11

Full Changelog: v0.3.22...v0.4.1

v0.4.1-alpha

Fully NATS integrated PoC, ready for stress testing.

v0.3.22

First pre-produciton version that is stable under load test. Key features:

• Improved LogDB implementation
• Improved distribution speed and memory efficiency with xxHash
• Full snapshot recovery, and backup support for nodes going down and coming up in cluster
• Automatic snapshot based recovery for nodes lagging far behind

v0.3.21

Return of SQLiteLogDB with quite a few bug fixes, which means lesser memory used and compacter LogDB compared to disk memory used with PebbleDB.

v0.3.19

For testing graceful transport failure and max log entries parameter

v0.3.18

Fully implementing save/restore sate. This will not be backwards compatible change, so please make sure you clear out your raft state directory (/tmp/raft by default). This version should allow you to run a cluster in a pretty stable shape, and even recover from cold start failures where a node was decommissioned for a while, coming back it will get a fresh new snapshot of database, and start the server.

v0.3.17

Fixed snapshot restore logic to reinstall hooks as well and not cause a crash with invalid DB image.