Often times when my replication breaks, it is a quasi monumental task to
Luckily, my replication is occurring on a data warehouse, where the
tables represent each days worth of data. If replication fails, I can
usually, get replication back on track by fixing whatever occurs in the
data/<servername>.err file. If the files get out of sync, I have to
abandon replication to get me back synced and reset the master "reset
master", remove the master and relay files on the slave.
Usually I will do a quick snapshot on the master and reset replication
on the master, transfer files to the slave, and start slave replication,
almost like doing a clean setup. Since I am out of sync by usually at
most 24 hours, I only have to transfer a couple of tables to get back
I guess the moral to this long winded story, is be very careful how you
design your tables so recovering from replication failure is easier.
Wells Fargo Services
From: Jeff Smelser [mailto:tradergt@stripped]
Sent: Thursday, April 14, 2005 1:16 PM
Subject: Re: slave out of sync
On Friday 08 April 2005 10:38 am, Chris Scheller wrote:
> i have replication going between 2 boxes. the master crashed a few
> days ago, and this morning i noticed that the slaves slave thread
> errored out. the binary log and offset had changed on the master and
> the slave couldn't sync up. i got the slave up and running again by
> changing the log file and offset in the slave. but now i am left with
> out of sync data. how do i get these back in sync, and how do i aviod
> this in the future?
Why wouldn't they sync up? If you left off at master-005, pos 123 (lets
It should pick up right there..
If your missing bin log files, then you need to set up the master to not
delete logs the slave has not processed yet. This would assure your up
to date, once the slave comes back alive..
Or am I missing something?