Validation errors in AWAITING_NEW_ROW.handleChunk() will always print rowKey=null and not rowKey=chunk.getRowKey

[joy91227]

rowMerger.push() will call stateMachine.consumeRow() for each chunk. After the completion of each chunk, stateMachine.reset() is called and state is set to AWAITING_NEW_ROW, and rowKey=null.

Since rowKey is null at the start of each AWAITING_NEW_ROW.handleChunk(), when we call Validate at L275, the rowKey that's printed as part of the error message will always be null.

java-bigtable/google-cloud-bigtable/src/main/java/com/google/cloud/bigtable/data/v2/stub/readrows/StateMachine.java

Line 275 in e11a6c3

validate(cmp < 0, "AWAITING_NEW_ROW: key must be strictly " + direction);

I think printing the rowKey in the error message is important and can speed up debugging.

For example, I have gotten the error message [1], and knowing the rowKey could help identify what went wrong.

[1]
Error message from worker: java.io.IOException: Failed to advance reader of source: BigtableSource{...}
…redacted………
Caused by: com.google.cloud.bigtable.data.v2.stub.readrows.StateMachine$InvalidInputException: AWAITING_NEW_ROW: key must be strictly increasing. numScannedNotifications: XX, numRowsCommitted:XX, numChunksProcessed: XX, numCellsInRow: 0, numCellsInLastRow: XX, rowKey: null, last5Keys: <these keys are printed correctly, I them redacted for PII concerns>