lni
lni
Provide callbacks on more raft states. Need to check with users to see what they need.
StopCluster should wait for the node to be fully unloaded. This will prevent two instances of the same node appear in the execution pipeline.
I've been trying all those exciting Paxos variants in the past few weeks. Will add some of them to dragonboat in H2, 2021 (towards the end of 2021). Yes, Dragonboat...
In the current implementation, dragonboat has to create periodic metadata-only snapshots (usually just a few kbytes each) for on-disk statemachines so they can be restored after reboot. This is largely...
need to coordinate with the engine to ensure such leadership transfer is requested before removing the node from nodehost.
Current version of dragonboat provides a tool called ImportSnapshot to repair clusters that permanently lost their quorums. For example, a 3 nodes cluster becomes unavailable when it loses 2 nodes...
Dragonboat currently uses TCP based transport for exchanging Raft messages between NodeHost instances. Such strictly ordered reliable transport is not actually required. The existing TCP based transport is pretty straight...
The chaos testing framework requires the ability to reorder messages sent between two nodes. Add interface to allow custom code to implement such reordering. The chaos testing framework will access...
run make static-check to see all those errors.