Fw: Re: Out of memory error with PG10.3, 10.4 but not 9.3.19

[ChatPristi]

 Forgotten to CC the list, sorry...

>Well, instead of an explain output which takes 2.4MB compressed and
>9.6MB uncompressed (take it as unreadable), could you produce a
>self-contained test case with a glimpse of the schema you are using?
>Where does the OOM happen, and how did you change your partitioned table
>schema?  Are you using the native partitioning instead?
>Michael,
Thank you for your answer.

Sorry for the unreadable explain output.
I attached a SQL dump with 2 entities loaded in the database (2,872,265 
entities in the actual database), the actual query and the actual output.

The OOM is durin the query (SELECT) after ~9 minutes the memory of the postgres 
increase until 8GB and the OOM message.

Partitioning is done by inherhitance.
After a complete reload of the database in PG10.4 the OOM still exists.


--
Michael<>

Fw: Re: Out of memory error with PG10.3, 10.4 but not 9.3.19

[ChatPristi]

 
Forgotten to CC the list too, sorry. again..
  >1073741818 is a bit less than 1GB and 1073741818+32 is a bit more. So
>you are obviously hitting a 1GB limit here.
>Given that 1GB is the maximum length of a character type value in
>PostgreSQL and the error message mentions a "string buffer", I suspect
>that your query tries to construct a very long string. Try to rewrite
>the query so that it creates several shorter strings instead.>      hp

Peter,
Thank you for your answer.I understood tht some buffer limit is reached. But I 
do not see where.The output is 1513 rows and 78 kb.


-- 
  _  | Peter J. Holzer    | we build much bigger, better disasters now
|_|_) |                    | because we have much more sophisticated
| |  | h...@hjp.at        | management tools.
__/  | http://www.hjp.at/ | -- Ross Anderson