On 1/16/2007 12:22 PM, Andrew Sullivan wrote:
> On Tue, Jan 16, 2007 at 09:19:06AM -0800, Alan Hodgson wrote:
>> > Restarting the slon processes cleared out old connections and brought
>> > back sanity.
>> 
>> I've seen this, too.  Restarting all the slons always clears it up.
> 
> It'd be really nice if people could start gathering as much forensic
> data on this issue as possible, since presumably it'd be nice not to
> have to do that all the time.

This symptom can not only be caused by the remote node not catching up. 
It can be perfectly fine in sync, but the sl_confirm entries aren't 
received back on the master, so the master "thinks" the slave is behind. 
One can simulate the problem by messing up the sl_path information 
required for the master receiving events from the slave (the same path 
is used to receive confirmations). Look for error messages in the slon 
logs where a master or provider node cannot connect to a subscriber.


Jan

-- 
#======================================================================#
# It's easier to get forgiveness for being wrong than for being right. #
# Let's break this rule - forgive me.                                  #
#================================================== JanWieck at Yahoo.com #