It would be convenient if fdbdr switch ensured that each cluster had a dr_agent pointing to the other cluster before locking the new destination cluster. When I was experimenting with it, I incorrectly assumed that the dr_agent on the original source cluster would be enough. I was confused when the fdbdr switch command hung and both clusters were locked.
Would you like to open an issue on foundationdb’s github [1] about this?
For sure. For reference, here’s the issue I just opened: https://github.com/apple/foundationdb/issues/1083
1 Like