Details
-
Improvement
-
Status: Open
-
Normal
-
Resolution: Unresolved
-
None
-
None
Description
We don't do power failure testing at the moment so we can't actually verify the correctness of Cassandra under those conditions. See for example the sstable synchronization problems for compressed metadata reported in CASSANDRA-10534.
We should perhaps use a mock file system that defers writes so we can simulate the kernel not flushing after the process exits.
Attachments
Issue Links
- relates to
-
CASSANDRA-10534 CompressionInfo not being fsynced on close
- Resolved