In 0.216.3 (Windows, SQLite DB, via npm, default EXECUTIONS_PROCESS value) we had a very simple workflow with a Kafka trigger that reads 1 kafka message and sends it to the Code node which runs code that takes about 10s to finish and prints start and end timestamps. If we write, for example, 5 messages into the Kafka topic, 5 parallel executions would be triggered, they print the start timestamp, and 10s later they print the end timestamp (they all finished roughly at the same time)
In 0.236.2 (Linux, Postgres DB, via npm, default EXECUTIONS_PROCESS value) we created the same workflow, but if we write 5 messages into the Kafka topic, the 5 workflow executions run sequentially: execution 1 for 10s, then execution 2 for 10s, and so on.
No error message, but we prefer the previous behaviour, as it can benefit from paralellization. Is this configurable? For us, this process went from ~10s to ~50s