We are running version 6.0.4.
There were 3 XDCR replications to another cluster of the same version since many months, and those were running fine. We added a 6 new XDCR replications to a new cluster of version 6.6 and it was fine for a few hours. But after a few hours, we see that memory usage of goxdcr process started increasing. It kept increasing for the next 10-12 hours, eventually eating up almost 70% memory on each node and then we had node & rebalance failures. After deleting the newly added replications (to cluster running version 6.6), memory utilisation returned to normal. We are not very sure if the node & rebalance failure was related to high memory utilisation but it looks to be so. To repeat, all this while, the replications to cluster with same version continued fine without any issues.
Has anybody encountered such an issue or if someone can provide more insights that it will be very helpful.