Обсуждение: permission denied

On 10/31/2013 12:53 AM, Pascal Tufenkji wrote:
> Hello,
>
> I am facing a weird permission problem with the views in our database.
>
> While the system is up and running, suddenly a certain view gets
> corrupted. When we try to query it, it returns “permission denied”.
> Although the permission has not been changed and if I check them using
> “\z” the system displays that all the permissions are still on. The only
> solution is to drop the view and create it back again.
>
> 3 views had been corrupted the past week.
>
> Until now, all the errors are on our SIT and UAT servers. I am afraid
> that soon we will be facing those same problems on our production database.
>
> We are using PostgreSQL 8.2.11 on Red Hat

I might as well be the first to say it, version 8.2 is no longer supported.

>
> What seems to be the problem ?

Hard to say.

>
> Can anyone help please ?

What is the exact error message returned?

How are the views being defined?

Are the tables under the views being changed?


>
> *Pascal TUFENKJI
> *Service de Technologie de l'Information
> Université Saint-Joseph - Rectorat
> Tel: +961 1 421 132
> Email: ptufenkji@usj.edu.lb <mailto:ptufenkji@usj.edu.lb>
>


--
Adrian Klaver
adrian.klaver@gmail.com

"Pascal Tufenkji" <ptufenkji@usj.edu.lb> writes:
> I am facing a weird permission problem with the views in our database.

> While the system is up and running, suddenly a certain view gets corrupted.
> When we try to query it, it returns �permission denied�. Although the
> permission has not been changed and if I check them using �\z� the system
> displays that all the permissions are still on. The only solution is to drop
> the view and create it back again.

That's pretty odd, but ...

> We are using PostgreSQL 8.2.11 on Red Hat

... you're not going to get a lot of interest in supporting a release series
that's been EOL for nearly two years, especially when you're a dozen minor
releases behind even in that series (the last 8.2 release was 8.2.23).
You are *way* overdue for an update.

If you can still reproduce this problem in a currently-supported release,
we'll definitely take an interest.

            regards, tom lane