Re: RLS + XPATH - Mailing list pgsql-hackers

From Joe Conway
Subject Re: RLS + XPATH
Date
Msg-id 115cc7ce-49fe-de0c-6ee3-2f6383604230@joeconway.com
Whole thread Raw
In response to RLS + XPATH  (Joshua Drake <jd@commandprompt.com>)
List pgsql-hackers
On 10/31/22 13:37, Joshua Drake wrote:
> Team,
> 
> While on the road in Iowa visiting covered bridges I met up with an 
> amazing individual named Brent. Brent, is with a small organization 
> named: Darpa.
> 
> They are using PostgreSQL + RLS + XPATH but unfortunately the 
> performance has been less than what those who use PostgreSQL on a daily 
> basis would expect. Specifically when using RLS + XPATH, the optimizer 
> does not prefilter and will scan every row. This has taken their 
> previous queries on another database from sub second, to 25+ seconds on 
> PostgreSQL.
> 
> Unfortunately, we were at a covered bridge park (see attached) and 
> weren't able to get deeper into the issue. That said, I did tell him 
> that I would report the issue. Maybe there is traction, maybe there 
> isn't to get this issue fixed.


The related functions need to be marked leakproof to get decent 
performance from RLS.

-- 
Joe Conway
PostgreSQL Contributors Team
RDS Open Source Databases
Amazon Web Services: https://aws.amazon.com




pgsql-hackers by date:

Previous
From: Ilya Gladyshev
Date:
Subject: Re: Segfault on logical replication to partitioned table with foreign children
Next
From: Ilya Gladyshev
Date:
Subject: Re: Segfault on logical replication to partitioned table with foreign children