On 1 October 2016 at 10:10, Tom Lane wrote:
> David Rowley writes:
>> On 1 October 2016 at 05:47, Tom Lane wrote:
>>> Somebody will need to trace through this on Windows and see where it's
>>> going off the rails.
>
>> I
David Rowley writes:
> On 1 October 2016 at 05:47, Tom Lane wrote:
>> Somebody will need to trace through this on Windows and see where it's
>> going off the rails.
> I tried the test case on 9.6.0 on a Windows 8.1 machine, and it works
> fine
On 1 October 2016 at 05:47, Tom Lane wrote:
> FWIW, I tried your original example on Linux yesterday, and it seemed to
> work fine --- it only wanted to use 1 worker, but the speedup was just
> about exactly 2X with or without "explain analyze". So this is somehow
> Windows
Jay Knight writes:
> I've tried this on a CentOS VM (VirtualBox on Windows) that I gave 2 cores,
> and it worked as expected (it launched workers with and without explain
> analyze), so I've only been able to reproduce this on Windows.
FWIW, I tried your original example on
I've upped max_worker_processes to 16, but I still can't get it to launch
workers unless I use EXPLAIN ANALYZE. I've also found that this simplified
setup exhibits the same behavior:
create table big as (
SELECT generate_series(1,3000) AS id
);
explain analyze SELECT avg(id) from big
On 30 September 2016 at 10:47, Jay Knight wrote:
>>What's max_worker_processes set to?
>
> 8
>
>>One theory would be that, the worker might not have been available
>>when you performed the query execution, but it just happened to be
>>when you did the EXPLAIN ANALYZE
>
> This
>What's max_worker_processes set to?
8
>One theory would be that, the worker might not have been available
>when you performed the query execution, but it just happened to be
>when you did the EXPLAIN ANALYZE
This happens consistently this way. Every time I run it with explain
analyze it uses
On 30 September 2016 at 08:52, Jay Knight wrote:
> So, why might postgres parallelize the query when I explain analyze it, but
> not when I just run it by itself?
One theory would be that, the worker might not have been available
when you performed the query execution, but it
Hello,
I installed 9.6 on Windows 7, and am experimenting with the new parallel
query feature. I've found a behavior that seems inconsistent to me.
Consider these two tables:
create table t1 as (
with r as (
SELECT generate_series(1,30) AS id, ceil(random()*25)::int as item,