[pgpool-general: 1020] Re: using prepared statements when memory_cache_enabled=on

[Lonni J Friedman]

On Wed, Sep 19, 2012 at 1:56 AM, Tatsuo Ishii <ishii at postgresql.org> wrote:
>> On Sat, Sep 15, 2012 at 6:05 PM, Tatsuo Ishii <ishii at postgresql.org> wrote:
>>> Lonni,
>>>
>>> Can you call pq_trace in your PHP app and send me the log when the
>>> error occurs?  It shows detailed data between your app and pgpool. It
>>> will help us lot to understand what's happening.
>
> I see C packet (Command Complete) immediately follows T(row
> description, response to "D"(Descrive) packet). This is strange.
> C should appear
> From backend> T
> :
> :
> From backend> C
> From backend (#4)> 13
> From backend> "SELECT 0"
>
> "C" should appear after "D"(Data row). But even strange thing is we
> see two "D" packets:
>
> From backend> D
> From backend (#4)> 286
> From backend> D
> From backend (#4)> 286
>
> This is the reason why you see the error ("D" should not appear
> without prior "T").
>
> I'm still looking for the cause of this.
>
> BTW do you see same error if you use memcached, rather than shmem?
> If not, I can narrow down the error case around the shmem code.

Actually we've been historically using memcached (for years), and
disabled it when trying to use the memcached support in pgpool.  We've
never had any problems with memcached alone (without pgpool's
memcached support).