-
Type:
Task
-
Resolution: Unresolved
-
Priority:
Major - P3
-
None
-
Affects Version/s: None
-
Component/s: None
-
Storage Engines - Server Integration
-
SESI - 2026-01-13
-
None
-
None
-
None
-
None
-
None
-
None
-
None
We need to create a metric to detect and track instances when a WT (WiredTiger) page read from the PageServer is corrupted.
Page corruption can occur at various points in the disaggregation architecture:
- Primary sending corrupted pages to the LogServer:
Pages can be corrupted before being sent to the LogServer. - Primary encryption errors:
Pages may be encrypted incorrectly by the primary node due to issues with the Key Encryption Key (KEK) or Data Encryption Key (DEK). - PageMaterializer corruption:
Pages may get inadvertently corrupted while being processed by the PageMaterializer. - PageServer corruption:
Pages may become corrupted within the PageServer itself.
Implementing this metric will enable us to monitor and promptly identify corruptions to minimize their operational effects.
- related to
-
SERVER-114527 [DS] Validate standby/replica handling corrupted oplogs
-
- Backlog
-