Paultagoras

> Would you consider also publishing a fat jar into releases, that would contain all the needed classes for this to work without users needing to fetch multiple modules separately?...

Thanks for letting us know!

It does, it also raises another question - have you noticed any differences between the tables on each cluster?

> Can we add a test for it? It's exactly-once, which we technically only test in the integration suite (because it uses Keeper)

> @Paultagoras i see one of our cloud tests is falling please take a look before It was failing due to timeout, after warming the instance it reran successfully -...

Hi @mlivirov ! So this seems like it would have a race condition, if there are multiple tasks running - is that not the case with the jdbc connector mentioned?

Hi @hekike - I'm closing this because we haven't heard anything in a while, but if it's still an issue please comment with the details @mzitnik was asking for and...

Hmm that's an interesting one - is this on a self-hosted instance?

Did some digging into the code with @mzitnik and discovered that the config options for retry/timeout seem to only apply to Source tasks rather than Sink tasks. You can see...

We're using the built-in mechanics so let's not implement something ourselves here.