Re: WAL Internals question - Mailing list pgsql-novice

From Fabrice Franquenk
Subject Re: WAL Internals question
Date
Msg-id 44AB8693.10700@bull.net
Whole thread Raw
In response to Re: WAL Internals question  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-novice
Tom Lane a écrit :
> Fabrice Franquenk <Fabrice.Franquenk@bull.net> writes:
>
>> Now i tried to set the bgwriter_lru_percent to 100% and
>> bgwriter_lru_maxpages to 1000 and
>> i did not spot any difference with the disk activities, cpu occupation
>> or anything else from
>>
>
> Hmm, are you sure your changes took effect?  To get the bgwriter's
> attention you need to make the change in postgresql.conf and then
> SIGHUP or restart the postmaster.  I'd have thought you could easily
> measure a difference given that much change ...
>
>             regards, tom lane
>
> ---------------------------(end of broadcast)---------------------------
> TIP 6: explain analyze is your friend
>
>
I made sure that all changes in postgresql.conf took effect. I restarted
the postmaster and i could not spot
any difference.
The values of  bgwriter_all_percent and bgwriter_all_maxpages are all
default (and i set checkpoints segments value to 250). Could there be
some kind of connection between these parameters and the bgwriter_lru_*
parameters?

by the way, was i correct describing the transaction handling mechanism ?

thanks for your help

regards,
            Fabrice Franquenk

pgsql-novice by date:

Previous
From: Christoph Della Valle
Date:
Subject: Re: copy recursive data
Next
From: Fabrice Franquenk
Date:
Subject: Do checkpoints flush all data from shared buffers ?