"H. Meyer" <slony-9999-exp at sblk.de> writes:
> I've just recently started to use slony-1 (thanks to all you guys who
> replied to my last message about 32bit<->64bit OS replication btw :o).
>
> Everything seems to be working perfectly at the moment.
> However, there are some strange entries in sl_log_1.
> i.e.
>
>> select * from _replication.sl_log_1 where log_xid=39903977;
>  log_origin | log_xid  | log_tableid | log_actionseq | log_cmdtype |
>        log_cmddata
> ------------+----------+-------------+---------------+-------------+---------------------------------
>           1 | 39903977 |          72 |        691449 | U           |
> id='2210825' where id='2210825'
>
> The column "id" is the primary key of that table. And as far as I know
> (wasn't able to verify that yet) none of our queries would ever do
> something like UPDATE ... SET id=X,... WHERE id=X.
>
> But even if there are such queries, why would slony-1 need to log that?
> No data was modified, so nothing needs to be replicated.
> Just curious :o)

You might have a trigger that runs on a subscriber every time there is
an update to this table (say, for cache invalidation or such).  Even
though the modification is "pretty null", the application can still
depend on this.

> And one other thing. I've read in the documentation that someone wrote a
> script to get slony-1 stats printed in mrtg. Has anything similar been
> done for munin (http://munin.projects.linpro.no/) yet? If not, I might
> be looking into that. Not quite sure which values would be most
> interesting to display though.
> st_lag_num_events and/or st_lag_time in sl_status seem interesting.
> Anything else?

It might be interesting to count SYNC events in sl_event, and
particularly, to count both the counts of SYNC and non-SYNC events.

But you have certainly hit the obviously interesting bits...
-- 
let name="cbbrowne" and tld="linuxfinances.info" in String.concat "@" [name;tld];;
http://www3.sympatico.ca/cbbrowne/multiplexor.html
Lisp Users:
Due to the holiday next Monday, there will be no garbage collection.