[pgpool-hackers: 2914] Re: Pgpool-II 3.8: allow to specify load balance weight ratio for load balance parameters
Bo Peng
pengbo at sraoss.co.jp
Mon Aug 6 09:25:51 JST 2018
Ishii-san,
I fixed the whitespace and compiler warnings.
Also I improved the documentation.
Could you check the new patch?
On Tue, 31 Jul 2018 09:21:01 +0900 (JST)
Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
> Thank you for the patch!
>
> I have looked into the patch and here are some feedbacks.
>
> 1) git apply makes complainings.
>
> t-ishii at localhost: git apply ~/load_balance_ratio_v1.diff
> /home/t-ishii/load_balance_ratio_v1.diff:101: trailing whitespace.
> <xref linkend="guc-database-redirect-preference-list">, when multiple database
> /home/t-ishii/load_balance_ratio_v1.diff:142: trailing whitespace.
> node for a particular database connection at a specified load balance ratio.
> /home/t-ishii/load_balance_ratio_v1.diff:237: trailing whitespace.
> <xref linkend="guc-database-redirect-preference-list">, when multiple database
> /home/t-ishii/load_balance_ratio_v1.diff:324: trailing whitespace.
> /*
> warning: 4 lines add whitespace errors.
>
> 2) Compiler warnings.
>
> In file included from protocol/child.c:60:0:
> protocol/child.c: In function ‘select_load_balancing_node’:
> protocol/child.c:1743:18: warning: too many arguments for format [-Wformat-extra-args]
> errdetail("app_name: %s index is %d dbnode is %s", app_name, index_app,
> ^
> ../src/include/utils/elog.h:195:14: note: in definition of macro ‘ereport_domain’
> errfinish rest; \
> ^
> protocol/child.c:1741:5: note: in expansion of macro ‘ereport’
> ereport(DEBUG1,
> ^
> protocol/child.c:1794:7: warning: ‘no_load_balance_node_id’ may be used uninitialized in this function [-Wmaybe-uninitialized]
> if (i == no_load_balance_node_id)
> ^
>
> 3) Subtle comment typos.
>
> * If the wight is bigger than random rate then send to suggested_node_id.
> * If the wight is less than random rate then choose load balance node from other nodes.
>
> "wight" should be "weight", perhaps?
>
> Best regards,
> --
> Tatsuo Ishii
> SRA OSS, Inc. Japan
> English: http://www.sraoss.co.jp/index_en.php
> Japanese:http://www.sraoss.co.jp
>
> From: Bo Peng <pengbo at sraoss.co.jp>
> Subject: [pgpool-hackers: 2897] Re: Pgpool-II 3.8: allow to specify load balance weight ratio for load balance parameters
> Date: Mon, 30 Jul 2018 11:49:12 +0900
> Message-ID: <20180730114912.41c0cb8237871fb5a62b8d46 at sraoss.co.jp>
>
> > Hi,
> >
> > I added a new feature to allow to specify load balance weight ratio for
> > database_redirect_preference_list and app_name_redirect_preference_list
> > parameters. Attached is the first patch.
> >
> > You can specify the list of "database-name:node id(ratio)" pairs to
> > send SELECT queries to a particular backend node for a particular
> > database connection at a specified load balance ratio.
> > This load balance ratio specifies a value between 0 and 1,
> > and the default is 1.0.
> >
> > For example, by specifying "test:1(0.5)",
> > Pgpool-II will redirect 50% SELECT queries to the backend node of ID 1
> > for the connection to "test" database. And the other 50% SELECT queries
> > will be sent to other backend nodes.
> >
> > Example:
> >
> > If you spacify like below:
> >
> > database_redirect_preference_list = 'postgres:primary,mydb[01]:1(0.3),mydb2:standby'
> >
> > It means,
> >
> > - Route all SELECT queries on postgres database to the primary backend node.
> >
> > - Route 30% SELECT queries on mydb0 or on mydb1 databases to backend node of ID.
> > The other 70% SELECT queries will be sent to other backend nodes.
> >
> > - Route all SELECT queries on mydb2 database to standby backend nodes.
> >
> >
> > The notion of app_name_redirect_preference_list is same as the database_redirect_preference_list.
> >
> >
> > On Thu, 14 Dec 2017 14:08:03 +0900 (JST)
> > Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
> >
> >> > Currently we can specify load balance condition by using,
> >> > database_redirect_preference_list, and
> >> > app_name_redirect_preference_list like:
> >> >
> >> > postgres:primary
> >> >
> >> > This means, if connected database is postgres, every SELECTs will be
> >> > redirected to primary.
> >> >
> >> > It would be nice if we could specify load balance ratio like:
> >> >
> >> > postgres:primary(0.3)
> >> >
> >> > which redirects 30% of SELECTs to primary (that is, other 70% of
> >> > SELECTs will be redirected to the load balance node). If the primary
> >> > node is also the load balance node, 100% of SELCTs will be redirected
> >> > to the primary.
> >> >
> >> > Other example is:
> >> >
> >> > postgres:primary(0.0)
> >> >
> >> > which never redirects SELECTs to the primary. This is useful if user
> >> > never wants to redirect SELECTs on the primary. Actually we cannot do
> >> > this for now if a primary node changes time to time.
> >>
> >> Sorry I was wrong. This is possible today too by
> >> "postgres:standby". What we cannot do today is
> >> "postgres:primary(0.3)".
> >>
> >> > postgres:mydb[0-4]:1(0.3)
> >> >
> >> > which redirects 30% of SELECTs to node 1 (that is, other 70% of
> >> > SELECTs will be redirected to the load balance node). If node 1 is
> >> > also the load balance node, 100% of SELCTs will be redirected to node
> >> > 1.
> >> >
> >> > I suppose the new specification overrides the current "weight"
> >> > parameters for each node.
> >> >
> >> > Comments and suggestions are welcome.
> >> > --
> >> > Tatsuo Ishii
> >> > SRA OSS, Inc. Japan
> >> > English: http://www.sraoss.co.jp/index_en.php
> >> > Japanese:http://www.sraoss.co.jp
> >> _______________________________________________
> >> pgpool-hackers mailing list
> >> pgpool-hackers at pgpool.net
> >> http://www.pgpool.net/mailman/listinfo/pgpool-hackers
> >>
> >
> >
> > --
> > Bo Peng <pengbo at sraoss.co.jp>
> > SRA OSS, Inc. Japan
--
Bo Peng <pengbo at sraoss.co.jp>
SRA OSS, Inc. Japan
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: load_balance_ratio_v2.diff
URL: <http://www.sraoss.jp/pipermail/pgpool-hackers/attachments/20180806/72bbe5ff/attachment-0001.ksh>
More information about the pgpool-hackers
mailing list