Re: BUG #3269: PSQL does not display error output

Thanks for the help Tom and others!

Nope, definitely did not intentionally redirect the STDERR..

> RHEL3 had SELinux. Do you have /usr/sbin/getenforce and if so what does
> it report?

NO SElinux on this box... (don't have getenforce)

On the same box I had 7.3.6 for 3 years running flawlessly.. Decided to upgrade
(had 8.2.3 running flawlessly on the staging/dev servers, and 8.1.8 on dev servers),
since the RPM was not available on the website for RHEL ES 3 (although the dir is there),
downloaded the source, compiled, installed (everything went smoothly), changed pg_hba.conf,
postgresql.conf and started the application.. Everything works just fine except for this
error reporting by psql.
If I connect from the remote server (using psql 8.2.3 or 8.1.8), everything works just fine. Also
if I connect from that server (using 8.2.4 psql) to a remote 8.2.3 server, the problem
exists.
All this was obvious from the problem, but just to make it clear that to me it seems to be psql issue..
Also I tried connecting to a remote 7.3.5, but got "FATAL: unsupported frontend protocol" - assume that fine
- not that much backwards compatible.

I would have provided you with access to the box to test, but unfortunately this is a production
server, and the nature of business does not allow me to. Is there any test/anything else that
I can do to provide further testing/input to you?

Thanks!

Regards,
bojan

On Fri, 11 May 2007, Tom Lane wrote:

> Bojan Jovanovic <bjovanovic(at)bjovanovic(dot)com> writes:
>> How would STDERR get redirected from psql?
>
> The usual way, like "psql 2>/dev/null", but if you didn't know that then
> it's unlikely you did it.
>
> I have seen symptoms roughly like this one with really ancient SELinux
> policies (the first draft of the policy tried to suppress direct writes
> on /dev/tty from any Postgres executable, IIRC :-(). But I didn't think
> RHEL3 had SELinux. Do you have /usr/sbin/getenforce and if so what does
> it report?
>
> regards, tom lane
>