The GitHub Actions job "Check License" on fluss.git/feat/lance-batch-flush-config has failed. Run started by GitHub user XuQianJin-Stars (triggered by XuQianJin-Stars).
Head commit for run: 65f3373673c987b61eda4f3cbfbcbfccc8fb0753 / forwardxu <[email protected]> [lake/lance] Add byte-based flush threshold to LanceLakeWriter Introduce a new 'max_bytes_per_batch' Lance table property (default 0 means disabled) that lets the tiering writer flush a batch as soon as its underlying Arrow off-heap allocation reaches the configured number of bytes, in addition to the existing 'batch_size' row-count threshold. Motivation: - With very wide rows the row-count threshold under-flushes, driving peak allocator memory too high. - With very narrow rows the row-count threshold over-flushes and produces many tiny fragments, hurting Lance read performance. Behavior: - 'batch_size' semantics are preserved (defaults to 512 rows). - 'max_bytes_per_batch' set to 0 keeps the historical behavior. - A negative value is rejected with an IllegalArgumentException. - The new threshold is checked using the Arrow field vectors' current buffer size for the accumulated row count, so small numeric batches and wide string/binary batches are both handled correctly. Tests: - New LanceConfigTest covering default / override / zero / negative / invalid parsing paths for both thresholds. - Existing LanceTieringTest and Arrow util tests continue to pass. Report URL: https://github.com/apache/fluss/actions/runs/28695965370 With regards, GitHub Actions via GitBox
