On Fri, Jan 16, 2009 at 5:07 PM, Christopher Browne
<cbbrowne at ca.afilias.info> wrote:
> "Troy Wolf" <troy at troywolf.com> writes:
>> The BUG, I believe, has something to do with Slony closing the file
>> handle or otherwise losing access to the file when he hits the wait
>> condition. There is nothing, in reality, wrong with that tmp file--no
>> other reason Slony is unable to write to it.
>
> I do recall that we found an bug of a similar *sort* of nature with
> log shipping where the slon was getting rather confused when there
> were multiple nodes as it only had one file handle for accessing the
> archive files, and [mumble, mumble, something about multithreading]
> led to that file handle getting lost.
>
> How many nodes have you got?  Just two?  Or more?
> --
> (reverse (concatenate 'string "gro.mca" "@" "enworbbc"))
> http://www3.sympatico.ca/cbbrowne/nonrdbms.html
> M$ is  for people  who want a  half-way implementation  of yesterday's
> ideas tomorrow.
>

Thanks for the reply!  I have 2 "normal" nodes. That is, node 1 is the
origin and node 2 is the subscriber.  Of course I also have a 3rd node
that is being replicated using logshipping. I say "of course" because
why would I have logshipping enabled unless I had at least a 3rd node.
This problem would occur regardless of the existence of that 3rd node
as long as node 2 is configured to generate the logship files.

You think this issue has possible come up before then....I wonder if
any work-around was developed. (Other than, put the database in
single-user mode! <g>)