[Bucardo-general] Pushdelta sync with a PG snapshot / restore fails to work bucardo 5.4.1

[Raghu Udiyar]

On Wed, Feb 15, 2017 at 6:59 PM, Greg Sabino Mullane <greg at endpoint.com>
wrote:

>
> What are you using to "snapshot"?
>

This is Amazon RDS, we used their snapshot feature to create and restore
snapshot to a new RDS DB ( which I assume must be using pg basebackups )


> > active |  SELECT * FROM bucardo.bucardo_delta_check($1,$2)
> > active |  SELECT * FROM bucardo.bucardo_delta_check($1,$2)
> > active |  SELECT * FROM bucardo.bucardo_delta_check($1,$2)
>
> The first step is fixing these. It looks like you have multiple
> Bucardo processes running, each of which is still trying to run
> bucardo_delta_check(). Kill all your Bucardo proceses, make sure
>

We realised, and canceled these queries. The delta table is pretty large
for this table, I guess that is the reason for delta check to be slow


> nothing is still running by checking pg_stat_activity, then
> add this to your ~/.bucardorc file:
>
> quick_delta_check=0
>
> Then restart Bucardo and see what happens.
>

Sync is working correctly after the above change.

So, the 'select distinct' is quicker than the delta check for some reason.

-- Raghu
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://mail.endcrypt.com/pipermail/bucardo-general/attachments/20170222/9698eb9f/attachment.html>