Normally, all end user traffic hits only the master. If the
master fails, you need to break mirroring(disable mirroring)(you
can do this via editing the surgemail.ini or via the web admin)
while you repair the old master.
When you bring up the old master, make it the slave and restart
mirroring. swap slave/master if you want after syncing is
I just did that. My slave had a hard drive failure, I disabled
mirroring. Rebuilt the slave. Restarted mirroring as per the
docs on their website. After resyncing, I swapped master/slave.
Broke mirroring(on purpose) and then rebuilt the other server(new
OS). Then restarted mirroring.
On 2020-08-15 11:32, Eric Vey wrote:
I've been using mirroring for a couple of months and I would
get this odd message telling me that things have been repaired
on the 15th. So I decided to find out what this was about.
Turns out I had this repair setting on both the master and
the slave, so I turned off the slave setting. But then I got to
wondering about this instruction:
This setting runs a nighly resync to keep the cluster in
sync. Maybe be resource intensive on a large system! This
should always be disabled during a failure as it could cause
messages loss when the master is re connected.
If the master has failed, how exactly would one get in to the
setting to turn this off? Only thing I can think of is to hope
that when the master came back up, it didn't have any network
access. And gosh, one more thing to remember before putting it