if you can make an script to lock the databases one by one, dump the locked
one and directly pipe it to slave server...
but you will have some problem on replication thread is this way unless
control the databases listed in replication thread to be replicated!
while dbx in $(DBsList)
lock all DB on Master and wait the slave to fully sync
stop slave replication
check the master position
unlock all locked db unless the dbx
dump dbx and directly pipe it on slave server!
all dbx to replication thread on slave and start slave thread
and wait to sync
## it's just an on the fly idea! maybe have some bugs...(but i hope not)
Ali R. Taleghani
On Mon, Sep 16, 2013 at 3:43 PM, nik600 <nik600@stripped> wrote:
> I'm looking at the fastes solution to re-sync a SLAVE node after a
> I have 300 databases with 80 GB of data, as i want to miminize the
> down-time on the master node i'd like to:
> 1) run a first rsync from master to slave
> 2) shut down the master
> 3) run a 2nd rsync
> 4) start the slave
> 5) start the master
> is this the correct way?
> can i simply lock all write command to the master at step 2 instead of
> shutting down it?