Re: xlogdump behaviour translating dropped relations

On 7/26/06, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>
> "Diogo Biazus" <diogob(at)gmail(dot)com> writes:
> > On 7/26/06, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> >> I'm not sure it should be making any attempt to translate anything.
> >> What makes you think the oids even refer to the current database?
>
> > I'm getting a new database connection based on the dbNode of the current
> > xlog record. And I expect that the user informed a connection to the
> backend
> > that originated the xlog files. Is this not going to work?
>
> No. I for one need this thing to analyze WAL dumps from databases
> I don't even have access to, let alone have up-and-running locally.
> It's really not going to be acceptable for the program to fail if it
> can't have access to a database matching what it sees in the log.
> The dropped-relation problem is just the first manifestation you happen
> to have run into of why that's an unacceptable restriction, but there
> are other reasons.
>

Agree, but this is only an option, and not even the default behavior.
So the program could still be used to analyze logs without access to the
database.
But if you do have access to the cluster it might be useful to translate
names if you are looking for some specific operation in order to recover a
previous database state.
And if it isn't what other functionality would be good to help users
determining a target xid for recovery?

--
Diogo Biazus - diogob(at)gmail(dot)com
Móvel Consultoria
http://www.movelinfo.com.br
http://www.postgresql.org.br