I have a database (699221). It contains of 1.8GB data (707710). I am doing a complex query. Which required to load a 80MB index (732287). I restarted Postgresql so the cache is empty and it has to read the table and index from disk. Which I understand is an expensive process. But what I don't understand is even I split the index into a different tablespace located on a completely separate disk (mounted on /hdd2) there is still a very long I/O wait time. That index is only thing exist on that disk. Any idea why? Or any way I can find out what it is waiting for? Thanks. (running DTrace tool kit iofile.d script to show I/O wait time by filename and process) bash-3.00# ./iofile.d Tracing... Hit Ctrl-C to end. ^C PID CMD TIME FILE 2379 postgres 23273 /export/space/postgres8.3/lib/plpgsql.so 2224 metacity 24403 /lib/libm.so.2 2379 postgres 32345 /export/space/pg_data/pg_data/data/base/699221/2619 2379 postgres 40992 /export/space/pg_data/pg_data/data/base/699221/2691 0 sched 82205 <none> 2379 postgres 273205 /export/space/postgres8.3/bin/postgres 2379 postgres 1092140 <none> 2379 postgres 59461649 /hdd2/indexes/699221/732287 (running DTrace tool kit iofildb.d script to show I/O bytes by filename and process) bash-3.00# ./iofileb.d Tracing... Hit Ctrl-C to end. ^C PID CMD KB FILE 2379 postgres 8256 /export/space/pg_data/pg_data/data/base/699221/699473 2379 postgres 87760 /hdd2/indexes/699221/732287 2379 postgres 832472 /export/space/pg_data/pg_data/data/base/699221/707710.1 2379 postgres 1048576 /export/space/pg_data/pg_data/data/base/699221/707710 -- John -- Sent via pgsql-performance mailing list (pgsql-performance@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-performance
