Hanks, Dan wrote:
>
> Hi folks,
>
> Slony 1.1.5, simple master-slave setup.
>
> I find myself in the situation where we had to bounce the master db,
> and I?m finding now that it apparently affected replication. There are
> currently over 8M rows in sl_log_1. Is it worth trying to let things
> catch up, and if so any pointers on how to speed that process up? I?m
> running the slon daemons at ?g 50.
>
> With a backlog of this size, would I be better to just start over?
>
> Thanks for any pointers,
>
> -- Dan
>
I suppose that depends somewhat on the nature of the backlog.

8M rows is quite a bit of traffic :-).

If you've only got some small multiple of 8M rows of data, then it would
probably be more efficient to let it catch up.

The killer question: Was the master's slon process down for a long time?
e.g. - Is there one giant SYNC that comprises most of those 8M rows?

If so, then -g 1 is plenty; you'll have one Really Big SYNC that
probably mandates keeping watchdogs shut off because it might well take
an hour to process that Big SYNC.