Oops! That was pretty irresponsible from me, not providing
that information! :-(
Anyway, it's postgres 7.2.3 running on a RedHat 7.3 Linux
system, on a dual Athlon CPU (presumably installed
properly -- i.e., with the SMP kernel), with 1GB of
memory.
I installed postgres from the source, and changed the
parameters related to the shared buffers; I put something
around 250MB dedicated to shared buffers.
At peak times, we estimate that the system may do
something around 10 to 20 SQL statements per second;
maybe 70% to 80% of those are insert statements. The
largest table has around a million records (we clean it
up weekly, since it would otherwise keep growing to
infinity).
Sorry for the lack of information in my initial post! I
hope this will increase the chances that someone will
be familiar with the problem and can guide me to find
a solution!
Thanks,
Carlos
--
Justin Clift wrote:
> Hi Carlos,
>
> Just to give everyone a bit more background information, which version
> of PostgreSQL are you having the troubles with, and which operating
> system is it running on?
>
>
> [...]
>
>>
>> I'm at a loss trying to figure out a strange error message
>> I'm getting (at an excessively high rate -- around 200 times
>> per day, out of a total of several hundred thousands statements
>> executed per day).
>>
>> The error message reads:
>>
>> "FATAL 1: Socket command option e unknown"
>>
>>
>> And the server closes the connection. Actually, the
>> complete error message contains several lines, and reads:
>>
>> FATAL 1: Socket command option e unknown
>> server closed the connection unexpectedly
>> This probably means that the server terminated abnormally
>> before or while processing the request.
>>
>>
>> Except that the server does not shutdown (more statements
>> continue to arrive, and they're executed).
>>
>> I checked the docs, but I can't figure out what may be
>> happening, so I'm at a loss at trying to determine what to
>> do. (the one thing I did notice is that it is related to
>> one specific statement -- however, it's not that the
>> statement is unconditionally wrong: it does execute
>> several thousand times per day, and it succeeds most of
>> the time)
>