This is in the service of a live migration, where server A has heavy performance and space demands, so I would like to keep down the amount of disk-space used by logs - let's say expire_logs_days = 10.
If I set server B, which is a slave of server A, to expire_logs_days = 30, will it then work to use a 20 day old snapshot (of B, for argument's sake) to make a new slave of B, called C, that will be able to catch up, or does the fact that server A has binlogs purged on the 10th day somehow break that scheme?
I am using row-based GTID replication on Percona 5.6, b.t.w.