You're right .
I will begin by migrate postgresql version.
postgresql is now to 8.3 version . A lot of performance inprovements
have been made in this last version .
In any case I must do a pg_dump/pg_restore.
So , Should I try directly 8.3 version or is it more cautious to stay in
8.1.x version ?
Scott Marlowe a écrit :
> On Dec 19, 2007 1:57 AM, olivier boissard <olivier(dot)boissard(at)cerene(dot)fr> wrote:
>> Scott Marlowe a écrit :
>>> On Dec 18, 2007 4:21 PM, olivier(dot)boissard(at)cerene(dot)fr
>>> <olivier(dot)boissard(at)cerene(dot)fr> wrote:
>>>> My purpose is to uses different query plan depending on queries
>>>> I got the following problem : some queries are taking too much time and
>>>> postgresql does not take the most relevant index in account.
>>>> I search on internet and found that no HINT function was available in
>>>> I tried to change indexes parameters in postgresql.conf but eache time I
>>>> set a param to OFF (example nested_loop ,seqscan, ...) , some queries
>>>> freeze database .
>>>> How can we force postgresql to use a plan ?
>>> That's not how we do things in postgresql land (usually).
>>> You should figure out WHY your queries are picking the wrong plan, and
>>> then see if you can get them to pick the right ones. If it's a query
>>> planner bug, you report it here, or the perform or general lists, and
>>> it gets fixed. Generic hints aren't likely to happen any time soon,
>>> although I do believe 8.3 is introducing function costing of some
>>> kind, which seems like a useful idea.
>>> But, back to fixing your slow queries.
>>> 1: Increase statistics targets on the guilty columns and reanalyze.
>>> 2: Run explain analyze select.... and post the output here.
>>> 3: ???
>>> 4: profit?
>> Thanks for reply
>> I will extract some queries.
>> I want to precise that I use postgresql 8.1.0 .
> First step would be to update to the latest 8.1.x version, whatever
> that is. There have been a LOT of fixes in the 8.1 branch since 8.1.0
> came out, and one might be a fix for your query problem. minor
> updates are strictly bug and security fixes, so they're not likely to
> cause any problems.
> ---------------------------(end of broadcast)---------------------------
> TIP 5: don't forget to increase your free space map settings
In response to
pgsql-admin by date
|Next:||From: Marcin||Date: 2007-12-19 22:33:31|
|Subject: problem with compilation of dynalloc2 segmentation fault|
|Previous:||From: Tom Lane||Date: 2007-12-19 17:04:10|
|Subject: Re: Scripting postgres commands |