[HDDS-3361] Change replication logic to use PersistedOpState

• Type: Sub-task
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 1.0.0
• Component/s: SCM
• Labels: pull-request-available
• Target Version/s: 1.0.0
• Assignee: Stephen O'Donnell Assign to me
• Reporter: Stephen O'Donnell
• Created: 08/Apr/20 17:00
• Updated: 09/Apr/20 12:06
• Resolved: 09/Apr/20 12:06

Description

In an earlier change, we decided that the Datanode would change the state of decommissioning / maintenance replicas and report the new state in its container report. Using that assumption, the logic in replication manager was changed to expect the ContainerReplica to contain the decom / maintenance state.

However in HDDS-2592 it was decided that the DN would simply store the operational state and report only the operation state in the heartbeat. This state is reported back to SCM and stored in the datanodeDetails instances.

This means that earlier assumption (ContainerReplica will contain the decom state) is no longer valid.

This Jira is to change the logic so replication decisions are based on the "persistedOpState" in the datanode details object instead.

Issue Links

• links to: GitHub Pull Request #789