Обсуждение: postgres-7.0 dying

Поиск
Список
Период
Сортировка

postgres-7.0 dying

От
hal Lynch
Дата:
Got a bit of a problem here?!
I just upgraded to Postgres-7.0
I am getting the error shown below consistantly, not always in
the same place in the database build script.
When I installed 7.0 I did not migrate my data.  Instead I chose
to rebuild all databases (6) and all tables (> 40) from scratch
from the raw data. This is a data warehouse so I just did a fresh
extract from the donor machine.  Let me say a word about the method
I use to populate the tables in the databases.  I extract the data
from the donor as tab delimited fields suitable for the psql 'copy'
command.  I then 'copy' the data into the table and build indexes.
This technique has worked for almost a year now with older versions
of Postgres.  The same exact data actually.

Hardware:
    pII 550
    lvd scsi drives (no file system is > 60% full)
Software:
    FreeBSD 3.4 (Postgres 6.5.3 has been running on this system
        for months without failure)
Postgres install:
    cd /usr/local/pgsql/src
    gmake
    gmake install
============================================================
The important part of the postmaster startup script
---------------------------------------
[ -x /usr/local/pgsql/bin/postmaster ] &&
    {
    #
    # BEWARE
    # The command exec'd by the su statement below uses
    # the default shell for user pgsql which is tcsh NOT sh.
    # That explains the syntax for the output redirect.
    #
    su -l pgsql -c 'exec /usr/local/pgsql/bin/postmaster \
        -i \
        -B 128 \
        -o "-S 1024" \
         -D /usr/local/pgsql/data \
            >& /var/postgresql/postgresql-7.0/startup.log'
    echo -n ' pgsql'
    }
==========================================================
The shell script which caused the error
---------------------------------------
#!/bin/sh

psql sis << \__EOD__
my_password
drop table stddmg;
__EOD__
rslt=$?

exit $rslt
==========================================================
The output of the above shell script
----------------------------------------
Password:
ERROR:  out of free buffers: time to abort !

ERROR:  out of free buffers: time to abort !

FATAL 2:  elog: error during error recovery, giving up!
pqReadData() -- backend closed the channel unexpectedly.
    This probably means the backend terminated abnormally
    before or while processing the request.
connection to server was lost
ERROR: drop table stddmg failed
========================================================
At the time I got this error the script had successfully
create 6 databases and about 40 tables;


Any clues for the clueless?????


Thanks

hal