Here is the scenario:
- Setup: 2 clusters with several unidirectional XDCR links. (From Cluster 1 to Cluster 2)
- Customer changes the Admin password on Cluster 2 (destination)
- No update of the Admin password on Cluster 1 (source).
- No XDCR on-going replications failed at password change time.
- Network outage experienced and all XDCR replications failed.
Due to security requirements, when there is a password change => all XDCR ongoing replications should be reset automatically. (not waiting for a pause/resume or a network outage)
We could imagine Cluster 2 (destination) doing a "drop all XDCR connections" which would force Cluster 1 (source) to re-initiate the on-going replications with the right password. Also to make it smooth and automatic, Cluster 1 (source) will probably need to support the old & new password.