Re: Help: massive parallel update to the same table

[rearranged - please don't top-post]

[also, bringing this back to the list - please keep the list copied]

Red Maple <redmapleleaf(at)gmail(dot)com> wrote:
> Kevin Grittner <Kevin(dot)Grittner(at)wicourts(dot)gov> wrote:

>> It should be parallel by default. Are you taking out any
>> explicit locks?

> my clients use psql to remotely run an update function on the
> postgres server. Each client run its own psql to connect to the
> server. What I have noticed is that if I commented out the update
> in the function so that only query is being done then all the core
> would kick in and run at 100%. However if I allow the update on
> the function then only one core would run.

> Currently it take 40min to update all the client statistics

Please show us the part you commented out to get the faster run
time, and the source code for the function you mentioned.

> Do you know if I have configured something incorrectly?
>
> I am running postgres 9.0.2 on fedora core 14. Here is my
> postgres.conf file
>
>
> [over 500 lines of configuration, mostly comments, wrapped]

If you're going to post that, please strip the comments or post the
results of this query:

http://wiki.postgresql.org/wiki/Server_Configuration

I don't think anything in your configuration will affect this
particular problem, but it seems likely that you could do some
overall tuning. If you want to do that, you should probably start a
new thread after this issue is sorted out.

-Kevin