[HDDS-8387] Container and volume scanners phase II

• Type: Improvement
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Component/s: Ozone Datanode
• Target Version/s: 2.0.0
• Assignee: Unassigned
• Reporter: Ethan Rose
• Created: 05/Apr/23 18:57
• Updated: 14/Mar/24 21:03

Description

Improvements to Ozone's container and volume scanner that are not required for the initial working implementation.

Issue Links

• split from: HDDS-7364 Improved container scanning

Sub-Tasks

• Container scanner continues to scan deleted container: Resolved
• CLI command to perform on-demand data scan of a specific container: Resolved
• Move RocksDB compaction out of disk checker: Resolved
• Improve container scanner metrics: In Progress
• Container export may fail for Schema v2 container while it is being scanned: Open
• Container delete may fail for Schema v2 container while it is being scanned: Open
• Provide CLI to do read-only scan of containers for debug purposes: Open
• Test that scanner does not impact performance: Open
• Use scanner to mark containers unhealthy on the write path: Open
• Add metrics for volume scanner: Open
• Allow on demand metadata scanning of open containers: Open
• Refactor volume scanner for easier synchronous checking: Open
• Allow container scanner to restore container health if health check passes: Open
• More graceful handling of recoverable volume failures: Open
• Inform on demand scanner if a specific block had an issue: Open
• Scanner to detect blocks on disk but not in DN RocksDB: Open
• Container scanner should warn about inconsistent metadata counters: Open
• Clarify locations of on-demand scanning: Open
• Schedule on demand scan of containers after import: Open