[Postgres-xl-developers] 答复: bug about parallel execution(Internet mail)

Tomas Vondra tomas.vondra at 2ndquadrant.com
Thu Jun 22 03:33:42 PDT 2017


OK, that seems like a good plan.

FWIW it may easily be a bug elsewhere, not necessarily in your patch - 
I've certainly seen cases where a path flag was not properly propagated, 
for example. And we're still missing a bunch of fixes from the 
PostgreSQL 9.6 branch (we've merged only up to commit 
b5bce6c1ec6061c8a4f730d927e162db7e2ce365, which is essentially 9.6alpha1).

BTW you seem to have responded to me directly, not to the mailing list. 
I do that mistake too sometimes, but please make sure you respond either 
to "all" or to "list" in the future.

regards
Tomas

On 6/22/17 3:39 AM, senhu(胡森) wrote:
> Hello:
>     Ok, I will make a check on my code and that patch to see why it does not work as expected.
> 
> Regards
> Senhu(senhu at tencent.com)
> 
> -----邮件原件-----
> 发件人: Tomas Vondra [mailto:tomas.vondra at 2ndquadrant.com]
> 发送时间: 2017年6月22日 4:18
> 收件人: senhu(胡森); postgres-xl-developers at lists.postgres-xl.org
> 主题: Re: [Postgres-xl-developers] bug about parallel execution(Internet mail)
> 
> Hello,
> 
> On 6/21/17 8:42 AM, senhu(胡森) wrote:
>> Hello, all:
>>
>>      I ran TPCH queries on PG-XL 9.6 a few days ago, and I have found
>> there is something wrong with parallel execution.
>>
>> Here is the SQL:
>   >
>> ...
>>
>> When I ran this sql, I got error message like this:
>>
>> ERROR:  Invalid Datanode number
>>
>> CONTEXT:  parallel worker
>>
>> I think the reason is that the parallel execution happened on
>> coordinators, not on datanodes. I have a patch to fix this, and the
>> patch is to make the parallel execution only happen on datanodes.
>>
>> We can have further discussion about this, and to find if there are
>> some more better sulotions about this.
>>
> 
> I agree we should disable parallel execution on coordinators, at least for now - and indeed that's what we tried to do in commit b6c7a69c. I'm not sure why it did not affect this particular plan.
> 
> We may allow parallelism on coordinators in the future, but at this point it may interfere with connection pooling for example (which I think might be the cause of your issue).
> 
> regards
> 

-- 
Tomas Vondra                  http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services


More information about the Postgres-xl-developers mailing list