Hi,

This seems to me a somewhat fundamental question that could be
immensely ful in understanding basic master slave replication:

If I delete a secondary index on a table on a slave db, should I
expect that replication will continue to work as normal? If I stop
replication, do a table optimization, then restart replication, would
I encounter any problems?

The understanding I'm coming to is that replication works by simply
playing back modifying queries recorded at the master on the slave, in
sequence. Such inserts/updates would be independent of secondary
indices, right?

Re cleanup: Let's say I have a bunch of old sessions, etc. that I
want to clean up - can I do this on the slave, assuming that the
application layer will either never query for these or that it will
simply create a new session?

If I could do this, seems reasonable that I could prepare a slave, get
it caught up, stop slave, do all kinds of aggressive maintenance,
start slave, get it caught up, then switch master and slave.

What about removing a completely unused all-null column? I'm guessing
that's where I could start to get into trouble...

Thanks in advance,


Marc