Hi all,
The database farm I've set up works great as a single master and two
slaves, provided nothing unusual happens to the master machine ... it
seems that if logrotate kicks in or the machine reboots, that the
mydbhost.001 file in /var/lib/mysql/ increments to 002, 003, etc., but
the slaves never update /var/lib/mysql/master.info to point at that
new file. Question 1: how do I fix that?
Question 2: the software I'm working with does a lot of writing to the
database and I'd obviously like to optimize that better than only
having a single writeable system.
I'm curious whether MySQL on the three machines would manage well if
the actual database files were mounted via NFS, or if there's some
better method to set up the replication.
A few of my Perl scripts have been modified to connect only to
"mydbhost1.mydomain.com" for making a $dbh_w handle, and to connect to
any of "mydbhost1", "mydbhost2" or "mydbhost3" for reading. But with
third-party software, rewrites to look for every SELECT call or every
INSERT/UPDATE/CREATE/DELETE call is obviously not going to work out
very well.
Any ideas or hints would be helpful.
-id