Well, if you're trying to run postgresql under openmosix, I'd suggest
looking into the archives of the postgresql-general and hackers mailing
list. note that this isn't a problem that has a fix in sight as far as I
know, i.e. there's a lot of work left to make it a functional solution,
and even the, it might wind up being slower than just running postgresql
on a single reasonably fast server.
If you are currently running your database on a mosix cluster I'd
recommend setting up a single machine and restoring your database to that
box.
As for getting your table back, I'm not so sure what to do. It sounds
like the pg_ catalogs are having some problems caused by being under
mosix, and you'll have to find someone here more familiar with the inner
workings of postgresql to get a fix.
On Wed, 17 Sep 2003, chakkara rangarajan wrote:
> Hello Mr.scott,
> Thx for the info and kind help. Can you please provide me any useful links
> about the nature of the problem.
> Rgds
> --Ranga
>
>
> -----Original Message-----
> From: scott.marlowe [mailto:scott.marlowe@ihs.com]
> Sent: Wednesday, September 17, 2003 10:26 AM
> To: chakkara rangarajan
> Cc: pgsql-admin@postgresql.org
>
> On Wed, 17 Sep 2003, chakkara rangarajan wrote:
>
> > Hi All,
> >
> > We have a development server running
> >
> > OS - Linux, 2.4.20-openmosix-r4
> >
> > DB - postgresql 7.3
> >
> >
> >
> > We have a table ctcert_name under postgres DB(postgres schema and postgres
> > user is the owner). Suddenly, this object started missing from the DB (I
> am
> > the only
> >
> > Person who connects to that server and did not drop/renamed it). When I
> > tried to recreate the same table, the system threw me back an error,
> saying
> > that "postgres.ctcert_name" already exists. I am neither able to drop or
> > rename the table.
> >
> >
> >
> > Can somebody please tell me, what cud have gone wrong and is the error is
> > reproduceable? What is the solution for this kind of problem.
> >
> > Your efforts are greatly appreciated.
>
> Postgresql is known to not work properly under Mosix. It may someday, but
> from what I've read the shared memory / semaphore system is at the heart
> of the problem.
>
>