Tom Lane wrote:
William Yu <wyu@xxxxxxxxxxx> writes:
If you have enough memory, you can always make a ramdisk to store the
WAL temporarily while you bulk load data. I also temporarily alter the
config to set checkpoint_segments = 1 to use as few pg_xlog files as
possible.
Really? That seems pretty counterproductive to me. I've never seen
a case where reducing checkpoint_segments increased performance.
Checkpoints are expensive, and usually you want 'em far apart.
Just poor grammar on my part. I meant decreasing checkpoint_segments
when using a ramdisk for WAL in order to keep Postgres from croaking on
lack of disk space for pg_xlog.
---------------------------(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
