Jay Levitt wrote:
> We need to do a few bulk updates as Rails migrations. We're a typical
> read-mostly web site, so at the moment, our checkpoint settings and
WAL are
> all default (3 segments, 5 min, 16MB), and updating a million rows
takes 10
> minutes due to all the checkpointing.
>
> We have no replication or hot standbys. As a consumer-web startup,
with no
> SLA, and not a huge database, and if we ever do have to recover from
> downtime it's ok if it takes longer.. is there a reason NOT to always
run
> with something like checkpoint_segments = 1000, as long as I leave the
> timeout at 5m?
There's nothing wrong with the idea except for the amount of WAL and a
huge checkpoint that can stall your system for a while in a worst-case
scenario. You can't get rid of checkpoint I/O completely.
I'd tune to a more conservative value, maybe 30 or at most 100 and see
if that solves your problem. Check statistics to see if checkpoints
are time-driven or not. As soon as almost all checkpoints are time-
driven, further raising of checkpoint_segments won't do anything for
you.
Yours,
Laurenz Albe
