Hello,
I have experienced an unlimitedly growing state in my streaming query in
table API and identified that the current implementation of time range
bounded over aggregate function was the cause. I was able to fix it by
modifying a couple of functions in flink-table-runtime-blink.
I am running several streaming applications in production and desiring this
fix to be merged to the official Flink. I have stated detailed issue
statements and fix plans in [FLINK-18119](
https://issues.apache.org/jira/browse/FLINK-18119) but didn't get reactions
for days. Please consider assigning the ticket.
Regards,
Hyeonseop
--
Hyeonseop Lee