[FLINK-9373] Fix potential data losing for RocksDBBackend

• Type: Bug
• Status: Closed
• Priority: Blocker
• Resolution: Fixed
• Affects Version/s: 1.5.0
• Fix Version/s: 1.5.0, 1.6.0
• Component/s: Runtime / State Backends
• Assignee: Sihua Zhou
• Reporter: Sihua Zhou
• Created: 16/May/18 06:48
• Updated: 06/Oct/21 07:50
• Resolved: 17/May/18 13:19

Description

Currently, when using RocksIterator we only use the iterator.isValid() to check whether we have reached the end of the iterator. But that is not enough, if we refer to RocksDB's wiki https://github.com/facebook/rocksdb/wiki/Iterator#error-handling we should find that even if iterator.isValid()=true, there may also exist some internal error. A safer way to use the RocksIterator is to always call the iterator.status() to check the internal error of RocksDB. There is a case from user email seems to lost data because of this http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/Missing-MapState-when-Timer-fires-after-restored-state-td20134.html

Issue Links

• is related to: FLINK-24460 Rocksdb Iterator Error Handling Improvement
• links to: GitHub Pull Request #6020