At Risk Tools
When a drive fails, all ring partitions that were on that drive immediately have only 2 copies left in the cluster (assuming 3 replicas here). As the replicators on the servers containing those other copies get to it, they will make an extra handoff copy to get back to 3 copies in the cluster.
Replication cycles can take a while though, so we'd like to have some tools to make this happen faster.
1) A tool that will list the ring partitions for a given device or a list of common ring partitions for a set of devices (for multi-device failures).
2) A tool to immediately start extra replication of a list of partitions from 1).
Blueprint information
- Status:
- Complete
- Approver:
- None
- Priority:
- Undefined
- Drafter:
- gholt
- Direction:
- Needs approval
- Assignee:
- None
- Definition:
- New
- Series goal:
- Accepted for grizzly
- Implementation:
- Implemented
- Milestone target:
- 1.7.5
- Started by
- gholt
- Completed by
- John Dickinson
Whiteboard
Part one was done and is in swift-ring-builder. Part two has yet to be done, but isn't being clamored for at the moment at least.