Details
-
Bug
-
Status: Resolved
-
Normal
-
Resolution: Fixed
-
None
-
Degradation - Resource Management
-
Low
-
Low Hanging Fruit
-
Code Inspection
-
All
-
None
-
[0e65e85861b84e87691d576ba80c5c3b3b7020d5|https://github.com/apache/cassandra/commit/0e65e85861b84e87691d576ba80c5c3b3b7020d5
-
Description
Active handling of FSError in AbstractLocalAwareExecutorService was introduced in CASSANDRA-14993.
However, in the current code base, there are places that catch the IOException but not wrap it with FSError, hence not triggering the DiskFailurePolicy. Especially when the bad disk no longer permits write operations, it could leads to the mutation stage backup. Therefore I propose to fix the IOException handling in those cases.
From the code inspection, those 6 places are current not re-throwing an IOException with FSWriteError.
- org.apache.cassandra.triggers.CustomClassLoader#addClassPath throws IOError. Invoked in TriggerExecutor's constructor
- org.apache.cassandra.io.util.FileUtils#renameWithConfirm throws RuntimeException
- org.apache.cassandra.io.util.FileUtils#truncate throws RuntimeException
- org.apache.cassandra.io.util.FileUtils#write throws RuntimeException
- org.apache.cassandra.db.compaction.LeveledManifest#sendBackToL0 throws RuntimeException. Invokes rewriteSSTableMetadata
- org.apache.cassandra.io.sstable.SSTableHeaderFix#writeNewMetadata throws RuntimeException. Invokes rewriteSSTableMetadata