Port Bug Report: Backend crashes daily.

============================================================================
POSTGRESQL BUG REPORT TEMPLATE
============================================================================

Your name : David Secret
Your email address : secret(at)kearneydev(dot)com

Category : runtime: back-end
Severity : serious

Summary: Backend crashes daily.

System Configuration
--------------------
Operating System : Linux 2.0.36 ELF/Redhat 5.2

PostgreSQL version : 6.4

Compiler used : gcc 2.7.2.3

Hardware:
---------
Linux tau.kearneydev.com 2.0.36 #1 Tue Oct 13 22:17:11 EDT 1998 i686 unknown
Pentium-II 266MHz 64M RAM

Versions of other tools:
------------------------
GNU Make version 3.76.1, by Richard Stallman and Roland McGrath.

--------------------------------------------------------------------------

Problem Description:
--------------------
At least once daily my visual basic 5 program using ODBC version
6.40.0004 kills the backend, all postmaster processes immediately
zombie and I'm dead in the water. There doesn't seem to be
a way to reproduce it regularly, as if I do the exact same thing
again it won't happen.

Here is an excerpt from the end of a -d 3 log:

<Normal queries & such for about 3 megs before this>
query: SELECT "material_id" ,"name" ,"short_name" ,"legacy" FROM "material" ORDE
R BY "legacy" DESC,"name"
ProcessQuery
CommitTransactionCommand
ERROR: postmaster: StreamConnection: accept: Invalid argument
pmdie 3
SignalChildren: sending signal 15 to process 11590
SignalChildren: sending signal 15 to process 11589
proc_exit(0) [#0]
shmem_exit(0) [#0]
.. <Everything dies from here>

I'll definately help by providing a sample of my program or
whatever debug level log files you'd like, I really love
PostgreSQL and will support it if I can use it in my projects,
except for large joins(7 tables) the performance is
impressive.

--------------------------------------------------------------------------

Test Case:
----------
Unfortunately I haven't found a way to reproduce the problem,
I think there is a small chance the ODBC driver is sending
something funky to the backend(as it mentions a failed
stream connect before dying), but no idea how to reproduce
this. I'll attempt to capture a log file from the ODBC
driver, but this is difficult as it only happens on 1 of 3
machines daily.

--------------------------------------------------------------------------

Solution:
---------

--------------------------------------------------------------------------