-
-
Notifications
You must be signed in to change notification settings - Fork 748
Open
Labels
Description
Bug fixes
Performance
- P2P blows up memory #8015
- P2P DataFrame performance - Disk overhead #7990
- Optional Disk for P2P shuffling #7572
- P2P shuffle is slow with string dtypes #7880
- Possibly fixed as of Improved conversion between
pyarrowandpandasin P2P shuffling #7896, needs to be evaluated.
- Possibly fixed as of Improved conversion between
- RFC Set priorities for p2p shuffle tasks #7926
- RFC P2P pass arrow tables directly to buffers #7992
Robustness
Diagnostics
Refactoring
- Restructure P2P code #8098
- Simplify state management in P2P's
ShuffleSchedulerPlugin#8018 - Make P2P shuffle extensible #8096
Documentation
- Improve user-facing documentation
- Add documentation for consistency/resilience mechanism
- Add documentation for developing your own P2P-based shuffle
- ...