Hi,
There is a design document for transaction support linked at the bottom of
KIP-98 that you can read here
https://docs.google.com/document/d/11Jqy_GjUGtdXJK94XGsEIK7CP1SnQGdp2eF0wSw9ra8
.
That describes some of the recovery/retry mechanisms. The design relies on
partition availability
to make
Hi, thanks for the answer, it was helpful.
So, if there are several topic-partitions in a transaction, the reads
are eventually consistent; it is possible some message from that
transaction are not available yet, until some recovery/retry mechanism
is completed for the fail topic-partitions? I
Hi,
This is very similar to traditional two-phase commit. There are essentially
multiple logs
being used - one per TopicPartition involved and the overall transaction log.
At the point
where COMMIT is being written to the TopicPartitions, it is assumed that it
will be possible to
write all of th
