Falko Menge

Vertical pod autoscaling and horizontal k8s cluster autoscaling can already be used today to dynamically control the hardware resources consumed depending on the workload.

Same stacktrace in 1.3.0-alpha3: ``` 2022-01-04 16:02:34.508 [Broker-0-StreamProcessor-1] [Broker-0-zb-actors-1] ERROR io.camunda.zeebe.broker.process - Unexpected error while processing resource '2 - Vacation Request Part4 - Trisotech-CamundaCloud3.bpmn' io.camunda.zeebe.model.bpmn.BpmnModelException: Collection expected to have entry...

Corresponding code seems to look for message subscriptions: https://github.com/camunda-cloud/zeebe/blob/develop/engine/src/main/java/io/camunda/zeebe/engine/processing/deployment/model/transformer/MessageTransformer.java#L39

However, the diagram only contains a message start event that shouldn't require a subscription, which would a correlation key: 

Okay, there are additional messages inside the XML that are not currently used. One could warn about them but this should not fail the deployment.

I removed the orphaned messaged using: ``` xmlstarlet edit --inplace --ps -N bpmn="http://www.omg.org/spec/BPMN/20100524/MODEL" --delete '//bpmn:message[not(//bpmn:messageEventDefinition/@messageRef=@id)]' *3.bpmn ``` but still get the error.

The issue is caused by an empty `bpmn:extensionElements` container, which I ended up with due to working around #8229 I worked around it by removing extensionElements without child nodes using:...

Thx @kpuljek! That fixed it for me too. Nevertheless, this should be automated by the uninstall scripts of Komorebi

Another refactoring suggestion: - [ ] [Smart Morph between Start, Intermediate, and End Events without loosing Event Trigger and Properties]( https://github.com/camunda/product-gaps/issues/73)

Using the JobClient should work as a workaround: https://github.com/camunda-community-hub/spring-zeebe#programmatically-completing-jobs