"Mikko Partio" <mpartio at gmail.com> writes:
> On 7/29/07, Jan Wieck <[[JanWieck at yahoo.com]]> wrote:
>
>           On 7/24/2007 12:29 AM, Mikko Partio wrote:
>      >
>      >     What I meant is that it'd be good to get enough detail information about
>      >     the specific actions that lead to this problem in order to create a
>      >     standalone test that can reproduce it.
>      >
>      >
>      > Ok here comes:
>      
>      This one played quite a bit hide and seek with me. But I think I found
>      and fixed the bug.
>      
>      The problem occurred when some action (like STORE_SET) caused the
>      subscriber slon to perform an internal restart. During that, the current
>      setsync tracking in the pset structure is reinitialized. However, since
>      the sl_setsync table is not updated on events other than SYNC, this
>      would lead to the wrong (too low) old sync expected if the last event(s)
>      processed from that node where no SYNC events.
>      
>      Thanks for the detailed example.
>      
>
>
> Great that you got it fixed! Will the patch be included in the 1.2.11 release?
> Regards
> MP

Yes, it's in what will be 1.2.11.

I have not been able to get the log shipping test in the
tests/testlogship directory in the code base to exercise the problem.
Now, the patch is applied in CVS, so that either a checkout of the 1.2
branch, or application of the patch to a copy of source that you may
have would allow you to deploy it for testing, at least.

I'd be more than happy to make a tarball available for you, if that
would help you run a test to verify that this fixes the problem for
you.
-- 
output = reverse("ofni.sesabatadxunil" "@" "enworbbc")
http://linuxdatabases.info/info/wp.html
"Consistency requires you to be as  ignorant today as  you were a year
ago."  -- Bernard Berenson