Re: Modifying Postgresql Optimizer to select optimal memory - Mailing list pgsql-general

From Tom Lane
Subject Re: Modifying Postgresql Optimizer to select optimal memory
Date
Msg-id 8421.1396392320@sss.pgh.pa.us
Whole thread Raw
In response to Modifying Postgresql Optimizer to select optimal memory  (Sayan Biswas <formal.sayan@gmail.com>)
List pgsql-general
Sayan Biswas <formal.sayan@gmail.com> writes:
> Hello everyone,
> We are currently working on a DBMS course project that requires us to
> modify postgresql optimizer. We need to vary work_mem within a range and
> note down the cost at those points(allocated memory) after 'explain'
> operation and then select the optimal tradeoff. We were changing
> "postgresql-9.1.12/src/backend/optimizer/plan/planner.c" where we created a
> new function mem_tradeoff() that calls planner() within a for loop for the
> varying memory. This mem_tradeoff() is called from pg_plan_query() in
> "postgresql-9.1.12/src/backend/tcop/postgres.c". The problem is, if the for
> loop runs only one iteration, everything is fine, but if we print
> plan->planTree->total_cost, it is printed 132 times for any query. Also if
> the for loop runs more than one iteration, we get a segmentation fault. Any
> help is very much appreciated.

The comment on standard_planner might help you:

 * Note to plugin authors: standard_planner() scribbles on its Query input,
 * so you'd better copy that data structure if you want to plan more than once.

            regards, tom lane


pgsql-general by date:

Previous
From: Stefan Keller
Date:
Subject: Re: Postgres as In-Memory Database?
Next
From: Tom Lane
Date:
Subject: Re: simple update query stuck