Ufuk Celebi created FLINK-5664:
----------------------------------
Summary: RocksDBBackend logging is noisy
Key: FLINK-5664
URL: https://issues.apache.org/jira/browse/FLINK-5664
Project: Flink
Issue Type: Improvement
Components: State Backends, Checkpointing
Reporter: Ufuk Celebi
Running a job with rocks db and a low checkpointing interval the logs are
flooded with
{code}
2017-01-26 17:29:39,345 INFO
org.apache.flink.contrib.streaming.state.RocksDBKeyedStateBackend -
Asynchronous RocksDB snapshot (File Stream Factory @
file:/Users/uce/Desktop/1-2-testing/fs/83889867a493a1dc80f6c588c071b679,
synchronous part) in thread Thread[Flat Map -> Sink: Unnamed (3/8),5,Flink Task
Threads] took 0 ms.
2017-01-26 17:29:39,398 INFO
org.apache.flink.contrib.streaming.state.RocksDBKeyedStateBackend -
Asynchronous RocksDB snapshot (File Stream Factory @
file:/Users/uce/Desktop/1-2-testing/fs/83889867a493a1dc80f6c588c071b679,
asynchronous part) in thread Thread[pool-88-thread-1,5,Flink Task Threads] took
53 ms.
{code}
Two log lines per stateful task per task manager per checkpoint. I can see that
this is helpful for debugging, but maybe too much by default. Should we
decrease the log level to debug?
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
