> What is the sustained volume of disk output you experience, for
> example from vmstat snapshots?
Unfortunately our metrics are... less than comprehensive at this point, so
I don't have data for that at the moment. Going to work to rectify, but
will take some time.
> 10% of 8GB (which is sounds
As pointed out you may want to look at some of your kernel parameters which
control the writeback (from cache) to disk.
Additionally you may want to take a look bg write parameters. Specially
bgwriter delay. I guess tuning that should help as well.
It might be useful if you can share your paramet
On Mon, Jun 16, 2014 at 11:01 AM, Nathaniel Talbott
wrote:
> We're still learning to tune PostgreSQL for our production setup, and over
> the weekend we saw a huge spike in INSERT/UPDATE completion time for a short
> period which turned out to correlate to an oversized checkpoint. I've
> included
