Thanks for the help but it was an SELinux policy problem.
I turned it off for now. I just do not have the time to learn SELinux
right now and my server is located behind a fire wall, contains not
sensitive data, and is just a test box for now.
Tom Lane wrote:
> David Siebert <david@eclipsecat.com> writes:
>
>>Okay I have version 8.0.2 installed on CentOS 4.0 with all updates.
>>I SSH into the server and try and run pg_dump all and get a stream of
>>errors on the server screen but not on my ssh.
>
>
> What do you get from
>
> ls -Z /usr/bin/pg_dumpall
> ls -Z /usr/bin/pg_dump
>
> ? On a reasonably up-to-date Fedora Core 3 system I get
>
> -rwxr-xr-x root root system_u:object_r:bin_t /usr/bin/pg_dumpall
> -rwxr-xr-x root root system_u:object_r:bin_t /usr/bin/psql
>
> but it sounds a lot like you have something else, like postgresql_exec_t
> (which is set up to forbid writes to /dev/tty, I believe). If so, try
> restorecon on these files to see if that fixes it. If not, you have an
> out-of-date SELinux policy RPM ... update that, or complain to CentOS
> that they haven't borrowed Red Hat's latest version yet ;-)
>
> regards, tom lane
>
>