since I moved from postgres 7.1 to 7.3 I get hangs when executing a
select on certaint rows with binary data using psql

The problem is reproducible and affects only single rows in the table .
The hangs happen only when I connect via network unsing psql -h
localhost template1

And the problem is gone, when ssl is switched off in postgresql.conf
But the problem seems to be on the client side, cause its the same using
a 7.1 server.

I have generated a testcase with 2 rows, one works the other not:

The table looks like this

create table testy (
id integer,
data bytea
);

(I can send a dump of the table with the offending rows to anybody
interessted by mail -- its about 130 K )

select * from testy where id=28162 --> works
select * from testy where id=28163 --> hangs + process goes up to
99% utilisation

The offending row is much larger, that might trigger.

I use a debian system with:
postgresql 7.3.3-1 client and server


--
Hans-Jürgen Hay <hjh (AT) alterras (DOT) de>


---------------------------(end of broadcast)---------------------------
TIP 4: Don't 'kill -9' the postmaster

=?ISO-8859-1?Q?Hans-J=FCrgen?= Hay <hjh (AT) alterras (DOT) de> writes:
We have heard a couple of reports of problems with SSL connections,
but so far I've not been able to reproduce any trouble here. Details
about your SSL installation might help.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match

Tom Lane wrote:

I pushed client_min_messages and log_min_messages to debug5, and the
only suspicious message at startup is

LOG: could not load root cert file "/usr/data/pgsql-7.4/root.crt": No
such file or directory
DETAIL: Will not verify client certificates.

which shouldn't be a problem (pg_hba.conf is configured to trust).

Any connect attempt will log
DEBUG: forked new backend, pid=1826 socket=8
DEBUG: proc_exit(0)
DEBUG: shmem_exit(0)
DEBUG: exit(0)
DEBUG: reaping dead processes
DEBUG: child process (pid 1826) exited with exit code 0

with the client side message "server does not support SSL, but SSL was
required". OpenSSL is 0.9.6g

Regards,
Andreas




---------------------------(end of broadcast)---------------------------
TIP 5: Have you checked our extensive FAQ?

http://www.postgresql.org/docs/faqs/FAQ.html

Andreas Pflug <pgadmin (AT) pse-consulting (DOT) de> writes:
Hm. The postmaster is sending back 'N' indicating that it does not want
to do SSL.

Are you sure you are connecting to an SSL-enabled postmaster?

Also, is your connection by any chance IPV6? It doesn't look like it
from the tcpdump, but I'm not sure I know how to tell.

The relevant bit of code in the postmaster is

#ifdef USE_SSL
/* No SSL when disabled or on Unix sockets */
if (!EnableSSL || port->laddr.addr.ss_family != AF_INET)
SSLok = 'N';
else
SSLok = 'S'; /* Support for SSL */
#else
SSLok = 'N'; /* No support for SSL */
#endif

so one of the paths that chooses 'N' is being followed. The test on
AF_INET is broken, since it should allow for IPV6, but otherwise I
don't see what's getting you.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend

Andreas Pflug <pgadmin (AT) pse-consulting (DOT) de> writes:
Yeah, I had just come to the same conclusion. Either you're connecting
to the wrong server (check those port numbers?) or there's something
wrong with the bit of postmaster code I just showed you.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend

Andreas Pflug <pgadmin (AT) pse-consulting (DOT) de> writes:
Right. I've committed the fix. One bug less, now to find the next one
....

Sounds good, see you then.

regards, tom lane

---------------------------(end of broadcast)---------------------------
TIP 8: explain analyze is your friend