[GitHub] [flink] StephanEwen opened a new pull request #10449: [FLINK-7289][state-backend] Integrate RocksDB memory use with managed memory

[GitBox <gi...@apache.org>]

StephanEwen opened a new pull request #10449: [FLINK-7289][state-backend] Integrate RocksDB memory use with managed memory
URL: https://github.com/apache/flink/pull/10449
 
 
   # What is the purpose of the change
   
   This PR sums up several PRs into a whole that allows users to either let RocksDB use Flink's managed memory budget, or to set a fix budget per slot which is shared between all tasks/operators.
   (The two are similar, in the first case the fix budget comes from the memory manager).
   
   ## Change log
   
   This PR is based on:
     - #10447 which adds shared resource tracking to the memory manager
     - #10416 which adds the use of a shared Cache and WriteBufferManager per slot
     - Some cleanups, simplifications, and integrations between the previous PRs.
   
   ## Follow-up
   
   This PR needs to be complemented with one further change: Releasing the handle to the shared resources when RocksDB is disposed. We plan to add this once #10300 is merged.
   
   ## Verifying this change
   
   This PR adds some unit tests.
   The PR can be manually verified by
     - setting Flink up with OFF_HEAP managed memory (it only works if all memory is OFF_HEAP, which is the only option once FLIP-49 is concluded)
     - Configuring RocksDB as the state backend
     - Setting in code (or in config) the use of managed memory (`state.backend.rocksdb.memory.managed: true`)
     - Enjoying a smooth running streaming program
     - Look for log lines like *"Getting managed memory shared cache for RocksDB."* and *"Obtained shared RocksDB cache of size {} bytes"*
   
   ## Does this pull request potentially affect one of the following parts:
   
     - Dependencies (does it add or upgrade a dependency): **no**
     - The public API, i.e., is any changed class annotated with `@Public(Evolving)`: **no**
     - The serializers: **no**
     - The runtime per-record code paths (performance sensitive): **no**
     - Anything that affects deployment or recovery: JobManager (and its components), Checkpointing, Yarn/Mesos, ZooKeeper: (yes / no / don't know)
     - The S3 file system connector: **no**
   
   ## Documentation
   
     - Does this pull request introduce a new feature? **yes**
     - If yes, how is the feature documented? **config options**
   

----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
 
For queries about this service, please contact Infrastructure at:
users@infra.apache.org


With regards,
Apache Git Services