Paŭlo Ebermann

``` nakadiClient .resources() .streamBuilder() .streamConfiguration( new StreamConfiguration() .subscriptionId(subscriptionId) .batchLimit(eventProcessingConfiguration.batchSize()) ) .streamObserverFactory(new EventStreamObserverProvider(processingService, eventParser, eventProcessingConfiguration)) .build() ``` The batch size here is 50, all other parameters are their default values.

We just got another case of this [(internal link)](https://jira.zalando.net/browse/LNG-547).

I guess a workaround would be to always set `max_uncommitted_events` to 1, but this will reduce the possible throughput quite a lot (no parallelization possible).

This would also allow to customize the generated threads, e.g. set a flow-ID.

Also, it would allow to not have the threads all named the same (`nakadi-java-compute-0` or `nakadi-java-io-0`). This makes it really confusing in the logs, if you have several StreamProcessors running...

@Jan-M I'm not talking about Grafana, just the normal alerting dashboards. (I'm one of the crazy people who is monitoring stuff for several teams.) It might happen for Grafana too,...

Additional idea: At the beginning of each chapter, add an edit link/button pointing to the corresponding .adoc-file on Github.

The second part now got a bit less urgent, as Github now can insert formatted HTML as markdown.

Is this still current? I guess the actual code changes need to be rebased onto current master, and the samples then regenerated.

Hi, thank you for your contribution. We don't think that this seemingly AI-generated image works well as an icon – it is a bit too information-dense. (Also, we like our...