> "Mark Woodward" <[EMAIL PROTECTED]> writes: >> psql -p 5435 -U pgsql -t -A -c "select client, item, rating, day from >> ratings order by client" netflix > netflix.txt > >> My question, it looks like the kernel killed psql, and not postmaster. > > Not too surprising. > >> Question, is this a bug in psql? > > It's really a libpq design issue: since libpq provides random access to > a PGresult, and has no mechanism for handling failures after returning > the PGresult to the client, it has to slurp the whole query result into > memory first. > > FWIW, there's a feature in CVS HEAD to instruct psql to try to use a > cursor to break up huge query results like this. For the moment I'd > suggest using COPY instead.
That's sort of what I was afraid off. I am trying to get 100 million records into a text file in a specific order. Sigh, I have to write a quick program to use a cursor. :-( ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match