Details
-
Sub-task
-
Status: Open
-
Major
-
Resolution: Unresolved
-
None
-
None
-
None
-
None
Description
The master branch currently has a framework for injecting corruption into an existing container and getting the expected result. As a part of reconciliation, the scope of this has expanded:
- We can detect more types of faults across replicas, including missing RocksDB metadata.
- The scanner can detect multiple different issues in the same container
- The expected result should now contain a merkle tree to compare against.
Previous PRs had used workaround solutions to test these things, but it would be good to have everything unified to ensure optimal test coverage and readability.
Attachments
Issue Links
- relates to
-
HDDS-10374 Make container scanner generate merkle trees during the scan
- In Progress
-
HDDS-11763 Implement container repair logic within datanodes
- Patch Available