ptupitsyn commented on code in PR #2215:
URL: https://github.com/apache/ignite-3/pull/2215#discussion_r1236880669
##########
modules/table/src/main/java/org/apache/ignite/internal/table/distributed/storage/InternalTableImpl.java:
##########
@@ -648,18 +648,26 @@ public CompletableFuture<Void>
upsertAll(Collection<BinaryRowEx> rows, InternalT
return enlistInTx(
rows,
tx,
- (commitPart, keyRows0, txo, groupId, term) ->
tableMessagesFactory.readWriteMultiRowReplicaRequest()
- .groupId(groupId)
- .commitPartitionId(commitPart)
- .binaryRowsBytes(serializeBinaryRows(keyRows0))
- .transactionId(txo.id())
- .term(term)
- .requestType(RequestType.RW_UPSERT_ALL)
- .timestampLong(clock.nowLong())
- .build(),
+ this::upsertAllInternal,
CompletableFuture::allOf);
}
+ /** {@inheritDoc} */
+ @Override
+ public CompletableFuture<Void> upsertAll(Collection<BinaryRowEx> rows, int
partition) {
+ InternalTransaction tx = txManager.begin();
+ TablePartitionId partGroupId = new TablePartitionId(tableId,
partition);
+
+ CompletableFuture<Void> fut = enlistWithRetry(
+ tx,
+ partition,
+ (commitPart, term) -> upsertAllInternal(commitPart, rows, tx,
partGroupId, term),
+ ATTEMPTS_TO_ENLIST_PARTITION
+ );
Review Comment:
There is no strong guarantee on the client. But embedded mode is
deterministic:
* We batch the data by partition, not by node
* Out-of-date partition assignment is not a concern
* Crashing nodes are not a concern
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
