Currently Redis replication is non-blocking from the point of view of the
master, but it is blocking from the point of view of the salve.
This can be a problem with some use case. For instance when there is a slave
farm in order to scale reads, if the master will go out the salves will lost
the link and will try to resync at the same time, even in a context where
serving old data is ok.
The solution is the ability to configure slaves so that they can act in
different ways while SYNC-ing with the master, for instance:
slave-sync-behavior error # Will report an error -ERR SYNC with master in
progress
slave-sync-behavior serve-old-data # Will continue serving old data while the
SYNC is in progress
Original issue reported on code.google.com by anti...@gmail.com on 24 Oct 2010 at 6:53
Original issue reported on code.google.com by
anti...@gmail.com
on 24 Oct 2010 at 6:53