[HDDS-2823] SCM HA support (implementing)
Authors: Li Cheng, Nandakumar Vadivelu, Rui Wang, Glen Geng, Shashikant Banerjee
2020-03-05
Summary
HA for Storage Container Manager using Ratis to replicate data
Abstract
Proposal to implement HA similar to the OM HA: Using Apache Ratis to propagate the
Links
The main SCM HA design doc is available from here
During the implementation of SCM-HA many smaller design docs are created specific to various areas:
- SCM HA Distributed Sequence ID Generator: about generating unique identifier across multiple nodes of the HA quorum
- SCM HA Service Manager: about starting and stopping the main SCM services (like PipelienManager, ReplicationManager) in case of a failover
- SCM HA SCMContext: about using a helper object which includes all the key information for all the required service components
- SCM HA Snapshots: about RAFT state-machine snapshots
- SCM HA: DeleteBlockLog: about coordinating block deletions in HA environment
- SCM HA: bootstrap: about initializing the SCM HA cluster