[Bucardo-general] PG async issue

[Chris Keane]

On Thu, Apr 17, 2014 at 4:38 PM, Chris Keane <chris.keane at zoomius.com>wrote:

>
>
> ...Also, the syncs from this server seem to now be glacial. I'm watching
> two identical clusters side by side, one with the freshly pulled HEAD and
> one with the 4.9.11 from a few weeks ago. An update to one database kicks
> all the right syncs in both systems and updates occur, on the older system
> in about 2 seconds. On the newer system, in about 3 minutes. The newer
> system also seems to get itself stuck after a while. The log shows the
> kicks are still arriving but after a while the updates get slower and
> slower until they stop happening. On a restart all the updates happen
> without issue. I may have to revert the pull until I can figure out what's
> going wrong.
>
> OK, so I dug into this a bit more. It looks like the notify in the kick
trigger changed? This is what it was before, and I've changed back to this
and all magically works again.

CREATE OR REPLACE FUNCTION bucardo.bucardo_kick_nhms_noncritical()
 RETURNS trigger
 LANGUAGE plpgsql
AS $function$
BEGIN
  EXECUTE $nn$NOTIFY bucardo, 'kick_sync_nhms_noncritical'$nn$;
RETURN NEW;
END;
$function$
;

The notifies just weren't getting through with the updated version (which I
didn't save for posterity). Database A = postgres 9.1,9, Database B =
postgres 9.3.2
The notify (and the kick) weren't making it from Database B back to
Database A.

Chris.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20140417/171c0aec/attachment.html>