Not GP, but early versions of Kafka had a lot of poorly-chosen default settings and sharp edges. eg, it was tuned for speed and availability over reliability, so it could lose your data if you weren't careful (not dissimilar to early versions of MongoDB that way). It was possible/easy to set up topics with no replication, so you'd lose data on your first server outage. KStreams had no real error handling - it would just explode and not even really tell you why. I wasn't as close to the operational side, but there were a lot of problems there too.
IME it has gotten better over time, but it's still pretty easy to shoot yourself in the foot if you don't understand what you're doing with it.