I don't think there's an automatic way to make the gateway senders on cluster 1 to re send all events to the gateway receivers on cluster 2. There are, either way, two options that come to my mind to avoid restarting member A:
- Use persistent regions on cluster 2. This way the members C and D will recover the data from disk on startup, and further updates from cluster 1 will be propagated as usual.
- Write a small function that iterates over the region and simply executes two operations for entry: get and put. By deploying this function to cluster 1 you can easily fire it whenever you want from GFSH, causing the events to be replicated to cluster 2 without needing to restart member A.
Hope this helps.