http://www.slony.info/bugzilla/show_bug.cgi?id=229

--- Comment #6 from Steve Singer <ssinger at ca.afilias.info> 2011-08-15 06:04:30 PDT ---
(In reply to comment #5)
> Unless the whole sequence is done in a try block, I am wondering if the
> automatic wait caused by switching event nodes is happening at the wrong moment
> here. It appears that it takes place AFTER the transaction for the "set add
> table" has taken out the lock.

The point of the patch is supposed to avoid that.

With this patch the intended behaviour is

1. Open a transaction against the new origin and lock sl_event_lock
2. Get ready to submit a SYNC event to the next origin. This triggers an 
   implicit wait for waiting on the old origin until it sees the sl_confirm.
3. Once the old origin is caught up the SYNC is submtted to the new origin.
4. The transaction commits
5. Open a new transaction getting ready for the set add table.

Steps 1-4 should ensure that everything is caught up before the transaction in
step 5 is opened.

Does my patch not do what I think? Or should I commit it

-- 
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.