Kyle Hanson wrote:

>There's a good chance you're correct about it have been being run on slave
>for last few months and not master, I'd say a 80-100 % chance of it.  What
>exactly does that tell me?
>  
>
If that's the case, then you'll have one really big SYNC to apply from
origin to subscriber.

In effect, if the origin's slon was down for all of February, and didn't
restart until March 4, then you'll find you have a Really Long SYNC; all
of the changes for February, and up to March 4, will be applied in one
swell foop as one big SYNC.

That SYNC could run for a potentially long time, and there's no way
(retroactively) to get any better behaviour.

This is why there is a script in the tools directory called
"generate_sync.sh"; its intent is that you run it every few minutes so
that even if slons are down, there will still be periodic SYNCs so that
you don't get stuck with loading a weekend's worth of data in one big SYNC.

Alas, you seem stuck with loading a rather bigger SYNC...