Stefan Pantos

Results 4 comments of Stefan Pantos

I've not been able to recreate what i've observed so i'm thinking that I've miss understood how commitAsync works. From what I've seen, if you do not provide a partition...

I've finally managed to recreate it in a horrible way. I added some hacky trace into the library to make sure I understood what was happening where and my checks...

There is a small mistaked in the code I gave but the problem still is apparent. As the committed offset is the last message consumed + 1 I should subtract...

I've created a PR to fix this incase it is helpful. https://github.com/confluentinc/librdkafka/pull/4553 I cannot think of a way to create a test for this however.