SQL command to dump the contents of table failed: PQendcopy() failed. Error message from server: socket not open

Hi:

I run a cron job every day to dump all the tables in my 7.4 Postgres
database. For one of the tables (sample) the command returns this error:

-bash-2.05b$ /usr/local/pgsql/bin/pg_dump -Fc -t sample heos -f
/home/bu/5/sample.dump
pg_dump: socket not open
pg_dump: SQL command to dump the contents of table "sample" failed:
PQendcopy() failed.
pg_dump: Error message from server: socket not open
pg_dump: The command was: COPY public.sample (field1, field2, field3,
field4, field5) TO stdout;

This is what the log shows:

-bash-2.05b$ more /usr/local/pgsql/logfile
LOG: database system was shut down at 2005-09-23 13:43:16 CEST
LOG: checkpoint record is at 122/E90B52AC
LOG: redo record is at 122/E90B52AC; undo record is at 0/0; shutdown TRUE
LOG: next transaction ID: 2227508; next OID: 30689327
LOG: database system is ready
LOG: server process (PID 27688) was terminated by signal 11
LOG: terminating any other active server processes
WARNING: terminating connection because of crash of another server process
DETAIL: The postmaster has commanded this server process to roll back
the current transaction and exit, because another server process exited
abnormally and possibly corrupted shared memory.
HINT: In a moment you should be able to reconnect to the database and
repeat your command.
LOG: all server processes terminated; reinitializing
LOG: database system was interrupted at 2005-09-23 13:54:37 CEST
LOG: checkpoint record is at 122/E90C3D38
LOG: redo record is at 122/E90C3D38; undo record is at 0/0; shutdown FALSE
LOG: next transaction ID: 2227617; next OID: 30697519
LOG: database system was not properly shut down; automatic recovery in
progress
LOG: record with zero length at 122/E90C3D78
LOG: redo is not required
LOG: database system is ready

This happens every day consistently. The table is quite big, about 1
million tuples and it is vacuumed.

Any help would be appreciated.

Cheers, Ruben.