http://www.slony.info/bugzilla/show_bug.cgi?id=235

--- Comment #11 from Christopher Browne <cbbrowne at ca.afilias.info> 2011-08-26 10:19:29 PDT ---
I have set up a branch for this that prototypes the proposed changes.

https://github.com/cbbrowne/slony1-engine/tree/bug235

There are a few issues outstanding at the moment:

1.  There's still a little question as to whether we start with grouping size
of 1, or if we go for something higher.

I'm assuming 1, at the moment, which has the merit that this is how Slony
already behaves.

2.  I'm logging the result of the computation at SLON_INFO level, which is not
the right verbosity level.

3.  There isn't a good test that validates the behaviour.  None of the
regression tests are really suitable, as they do not impose any big backlogs.

What is needed is a test that:
a) Loads a fair bit of data into some tables, so that subscription takes a
while;
b) After the subscription request is submitted, injects updates for the
duration of the subscription, so that there are backlogged updates during the
subscription.

If one of the disorder tests already has this pattern, then it would be useful
to know which it is so that we might observe the slon logs during that, to
verify that we're getting sensible behaviour.

-- 
Configure bugmail: http://www.slony.info/bugzilla/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.
You are the assignee for the bug.