On Thu, 2008-02-14 at 19:45 +0000, Christopher Browne wrote:
> Ow Mun Heng wrote:
> > Is there a way to manually force slony to switch from log_1 to log_2?
> > The table is getting way big (~1+G) and I think that it's causing the
> > "fetch 100 from log" to become slow ~60-150secs.
> >
> You can request this from SQL prompt via:
>
> select _my_slony_schema.logswitch_start();
hmxmms=# select _hmxmms_my_cluster.logswitch_start();
NOTICE: Slony-I: Logswitch to sl_log_2 initiated
logswitch_start
-----------------
2
(1 row)
Time: 1149.556 ms
cool, it has switched.
>
> It may fail with the error:
> ERROR: Previous logswitch still in progress
>
> It will not immediately lead to cleaning out sl_log_1; that is handled
> as part of the cleanup thread, which normally runs every 10 minutes
> (controlled by a parameter compiled into slon).
I've put this onthe CLI option slon -c2
I'm using version 1.2
>
> In versions 1.0 thru 1.2, further control of this takes place within the
> cleanup thread, which is almost exclusively C code in
> src/slon/cleanup_thread.c.
>
> In CVS HEAD, nearly all of this code has been folded out into a pl/pgsql
> stored procedure, cleanupevent(interval, boolean), which would allow a
> script to invoke cleanup.
Nice to know, but as of right now, way over my head.
Thanks
_______________________________________________
Slony1-general mailing list
[email protected]
http://lists.slony.info/mailman/listinfo/slony1-general
