Re: How exactly does Idle-state retention policy work?

Hi Srikanth, As you side, the idle-state retention time configuration only works for "group aggregate" and "over aggregate". Regarding to your question: > If it happens to be that the 'ip' acts as a key in my query for the eviction > to work, how does Flink justify the Heap size grew to 80GB an

How exactly does Idle-state retention policy work?

Hi there, I'm using FlinkSQL to solve to do the job for me. Based on this , configured the idle-state milliseconds. *context*: FlinkSQL reads Kafka stream with no key and put to dynamic table