-
Committer:
Kenny Gryp
-
Date:
2011-04-27 13:54:29 UTC
-
Revision ID:
gryp@dakin.be-20110427135429-w0tvdueljz22i73j
This patch fixes some bugs and adds some 'functionality' (read fixes bugs)
- don't do MASTER_POS_WAIT(newmasterfile and io) on the slave when it's replication is not properly running, it would hang forever
- slaves could become inconsistent when the new master is lagging on the slave
- remove the slave if it's not able to move the slave to the new master properly, and display an error. It used to just change master to a 'random' file and position
Note that this can make the process slower as it will let the new master sync and/or the slave first before changing the replication