Make tests access sandbox servers

[GoogleCodeExporter]

If you do

/etc/init.d/mysql stop

and then run all the tests, anything that fails needs to be fixed.  That's
the easiest way to wean the test suite off the localhost server :)

Original issue reported on code.google.com by baron.schwartz on 18 Oct 2008 at 9:17

[GoogleCodeExporter]

I agree (same w/ issue 104). In fact, I had the idea some weeks ago to create a 
much
faster, "universal" sandbox. What annoys me is that for each test the 
sandbox(es) are
stop/started and this is slow. Specially when I'm developing, I'm running a 
script's
tests over and over. So I thought: wonder if we setup the sandbox once, then had
other scripts/commands to simply reset tables or whatnot. It's a lot faster to 
drop a
table than restart the server.

We'd need then to develop some guidelines concerning how to play nicely with 
others;
i.e. how to clean up after yourself or not, etc. Myself, I have tests 
using/relying
on previous tests. I usually document such cases.

Then, we could design a complete sandbox environment: a master configed in such 
and
such a way, 2 slaves, standardized port numbers and user accounts for testing 
other
little things like when a user does have privs to do whatever (because currently
msandbox/msandbox has all privs).

We could even make a Test module of some kind to provide a unified interface to
making connections to master, slaves, loading sample files, cleaning 
up/resetting the
sandbox, etc.

Original comment by dan...@percona.com on 19 Oct 2008 at 4:33

[GoogleCodeExporter]

Original comment by dan...@percona.com on 30 Nov 2008 at 5:16

• Added labels: Milestone-2008_11

[GoogleCodeExporter]

Target these for release end of January 2009.

Original comment by baron.schwartz on 4 Jan 2009 at 5:04

• Added labels: Milestone-2009_01

[GoogleCodeExporter]

This is done as of r2801 and r2821.

Remember, some scripts require that you manually load the sakila db into the
persistent sandbox. If not, those test should skip the tests which require 
sakila.

Original comment by dan...@percona.com on 16 Jan 2009 at 8:25

• Changed state: Fixed

[GoogleCodeExporter]

Not fixed yet. Some scripts just assume a local MySQL is there, so grepping them
didn't show any access to old sandbox stuff, like mk-parallel-restore.t.

Original comment by dan...@percona.com on 17 Jan 2009 at 6:24

• Changed state: Started

[GoogleCodeExporter]

Now everything is fixed as of r2848.

Original comment by dan...@percona.com on 17 Jan 2009 at 11:13

• Changed state: Fixed