What's the generally accepted method for killing processes that went 'all wacky' in postgres?
I think i've seen in this group that kill -INT would be the way to go.
I'm playing around with different options for a median function. this one got out of hand
and was taking too long, so i wanted to kill it:
test=# select array_median(array(select t1 from test2 order by 1));
^CCancel request sent
It just sits there, it's been trying to die for 1/2 an hour.
At the OS it's taking up 100% of the CPU.
I tried kill -INT <pid> but that didn't help.
It's not updating anything, and i'm the only one in the database.
Fortunatly it's not production, so I don't really care. But if it was production, what would
be the method to kill it? (I know about kill -9, i'm assuming that == bad)
If this were production, I'd need to end the process, force a rollback (if necessary) and get
my CPU back so "just waiting for it to die" really isn't an option...
(PostgreSQL 8.3.5, linux/SLES11)
Thanks
Dave