I did a CLUSTER - it took just under 12 hours. The original load of the
table (including parsing all the geometries) took about 4 hours (+4 to
build the indexes and vacuum analyse). I dont know why the cluster is
so much slower...
My "SELECT * FROM <table> LIMIT 1;" is now fast.
Thanks for the help!
As a side note, I think the behavior of the Database is a bit
counter-intuitive. If I were to do a:
UPDATE <table> SET i = i + 1;
My table will "leak" 10Gb of disk space, make queries extreamly slow,
and require taking the database off-line for 12 hours to fix it!
Both the VACUUM and the UPDATE manual should be updated to say that this
is whats happening. If I do my above query 100 times, will I be leaking
a Terrabyte of disk space?
Will increasing my max_fsm_pages to 2,000,000 solve my problem?
dave