Jacob Yocom-Piatt wrote:
have two machines that are carped and have mysql databases on them.
one machine is the mysql master and the other a slave that replicates
the master. this keeps things ready for what i've just experienced:
mysql master machine goes down. now that a master failure has
occurred, i am not certain of the best path for bringing the master
back online so that it's synced with the slave.
to bring the master back online correctly i see the following as an
obvious, but laborious, path: (1) restart master and increase advskew
on the carp interface so the slave stays MASTER (2) edit config files
and restart mysql on both machines so that the slave is now the master
and vice versa for the original master (3) initiate slave replication
onto the original master (4 - optional) decrease advskew on the
original master machine so that it takes over (5 - optional) restart
mysql on both machines, swapping the master and slave configs. the
optional steps are only if restoring the original master as master is
desired.
if there is another way to do what i've described above that doesn't
involve downtime for mysql or is more easily automated, please advise.
in case anybody else encounters a similar situation, a friendly bird
whistled in my ear:
http://www.onlamp.com/pub/a/onlamp/2006/04/20/advanced-mysql-replication.html
cheers,
jake
