Re: Cache plan invalidation - Mailing list pgsql-hackers

From Tom Lane
Subject Re: Cache plan invalidation
Date
Msg-id 19707.1178378479@sss.pgh.pa.us
Whole thread Raw
In response to Re: Cache plan invalidation  (Bruce Momjian <bruce@momjian.us>)
Responses Re: Cache plan invalidation  (Bruce Momjian <bruce@momjian.us>)
List pgsql-hackers
Bruce Momjian <bruce@momjian.us> writes:
> Bruce Momjian wrote:
>> The current TODO list has:
>> 
>> Dependency Checking
>> ===================
>> 
>> * Flush cached query plans when the dependent objects change,
>> when the cardinality of parameters changes dramatically, or
>> when new ANALYZE statistics are available
>> 
>> A more complex solution would be to save multiple plans for different
>> cardinality and use the appropriate plan based on the EXECUTE values.

This is partially done --- you'll have to split it into multiple items
if you want to preserve the bit about keeping different plans for
different parameter values.  Note that in the current code, any VACUUM
or ANALYZE on a table will force relcache inval and hence replan; see
vac_update_relstats.  So the only case not covered as far as
non-parameterized queries go is large growth of a table without any
vacuuming or analyzing ... and you're going to have problems anyway
if you don't analyze after loading a table.  We may in fact find that
our problem is now too many replans rather than too few.

>> * Track dependencies in function bodies and recompile/invalidate
>> 
>> This is particularly important for references to temporary tables
>> in PL/PgSQL because PL/PgSQL caches query plans.

This is done.

> Also, is this done:

>  * Invalidate prepared queries, like INSERT, when the table definition
>    is altered

This too.
        regards, tom lane


pgsql-hackers by date:

Previous
From: "Jaime Casanova"
Date:
Subject: Re: Patch Status in the wiki
Next
From: Andrew Dunstan
Date:
Subject: Re: array type name mangling