As discussed with email@example.com on Gitter, we are having the issue where we have stuck transactions on one of our databases that causes Debezium to refuse to write its offsets, which results into the following messages being logged thousands of times every second:
This was after a fresh snapshot of the database. There are two transactions that have been open for well over a week now. Obviously they should be killed, but we have no control over that.
My understanding was is that there is currently no way to ignore transactions so this would effectively grind the CDC process to a halt since the offsets are not being committed to Kafka. Please correct me if I'm wrong
I've attached a trace log of the whole thing on his request of Chris (712k log lines, so it's a big one!)