Tom Lane wrote:
> Kevin Houle <kjh@cert.org> writes:
>
>>Is it just me, or are both sides reading waiting for the other
>>side to send data?
>
> Sure looks like it. Could it be an OpenSSL bug?
Well, redhat-9 ships with openssl-0.9.7a so I tried it
with openssl-0.9.7b and lo_read still caused a read()
loop after 32768 bytes. Then I used a Debian sid machine
for testing using the same test case. It uses
postgresql 7.3.3
openssl 0.9.7b
The lo_read worked flawlessly on the Debian box. So I grabbed
the postgresql 7.3.3 packages from the PGDG website and tried
those on redhat-9. Still experienced the loop after 32768 bytes.
I'm at a bit of a loss here. It would be nice if someone else
could run the test script (see original message in this thread)
on a redhat-9 box to eliminate my environment from the equation.
Kevin