Definetely it's a good start but currently we're building our disaster/recovery processes and some topic contents are really important. So in a paranoiac mode we want to some guarantees that messages are not corrupted. Are there some internals in storage mechanisms during the send and the replication which mitigate this kind of problem (I don't mean acknowledgement processes but for data themself)?
In the worst case, we could never send a message and duplicate an other and so have the same watermark. I guess that by design MirrorMaker v2 can't have this use case but we are paranoiac :) Le lun. 4 nov. 2019 à 23:58, M. Manna <[email protected]> a écrit : > How about High watermark check ? > > Since consumers consume based on HWM, presence of the same HWM should be a > good checkpoint, no? > > Regards, > > On Mon, 4 Nov 2019 at 22:53, Guillaume Arnaud <[email protected]> wrote: > > > Hi, > > > > I would like to compare the messages of an original topic with a mirrored > > topic in an other cluster to be sure that the content is the same. > > > > I see that checksum method in KafkaConsumer is deprecated: > > > > > https://kafka.apache.org/10/javadoc/index.html?org/apache/kafka/clients/consumer/KafkaConsumer.html > > > > What is the best way to do that ? > > > > Thank you, > > > > Guillaume > > >
