[Pgpool-general] pgpool-II 2.2 beta2 released

[Tatsuo Ishii]

As stated in the comment of the source, current solution is far from
an ideal one. We don't need to kill all the children, instead we could
kill processes which are using the database to be dropped.
--
Tatsuo Ishii
SRA OSS, Inc. Japan

> Ah ok I had not noticed that before.
> Thanks
> 
> Sent from my iPhone
> 
> On Feb 10, 2009, at 19:07, Tatsuo Ishii <ishii at sraoss.co.jp> wrote:
> 
> > The begavior you are seeing with DROP DATABASE is, actually exactly
> > same as all version of pgpool-II. To process DROP DATABASE, postgres
> > requires that there's no connection to the database. To satify the
> > requirement in very wild way: pgpool child sends signal to parent
> > pgpool and it does a restarting all child pgpool processes. See around
> > line 293 in pool_proto_modules.c.
> > --
> > Tatsuo Ishii
> > SRA OSS, Inc. Japan
> >
> >> Tatsuo,
> >>
> >> I'm starting to test some postgresql commands and I just saw this,
> >> besides commenting out the T_CreatedbStmt line in the
> >> pool_process_query.c file I also commented out T_DropdbStmt since the
> >> DROP database was also failing. So when I try to drop the database
> >> (although it works) pgpool log show the following:
> >>
> >>
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 23611: statement2: DROP DATABASE hello ;
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 23611: Query: sending SIGUSR1 signal to parent
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 32543: failover_handler called
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 32546: child receives close connection request
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 32547: child receives close connection request
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 32548: child receives close connection request
> >> Feb 10 22:46:37 set2_pgpool pgpool: 2009-02-10 22:46:37 DEBUG: pid
> >> 32549: child receives close connection request
> >>
> >>
> >>
> >> Marcelo
> >> PostgreSQL DBA
> >> Linux/Solaris System Administrator
> >>
> >> On Feb 9, 2009, at 4:21 AM, Tatsuo Ishii wrote:
> >>
> >>> From: Łukasz Jagiełło <lukasz.jagiello at gforces.pl>
> >>> Subject: Re: [Pgpool-general] pgpool-II 2.2 beta2 released
> >>> Date: Mon, 9 Feb 2009 10:27:08 +0100
> >>> Message-ID: <de899fb00902090127l121b0557j4234fe10c7648e93 at mail.gmail.com
> >>>>
> >>>
> >>>> 2009/2/7 Tatsuo Ishii <ishii at sraoss.co.jp>:
> >>>>> * pool_process_query.c: Start internal transaction if necessary.
> >>>>> Before we only started transaction INSERT, UPDATE, DELETE and
> >>>>> SELECT. Obviously we need to start transaction other than these,  
> >>>>> for
> >>>>> example CREATE TABLE.  To avoid an overhead to look up large  
> >>>>> number
> >>>>> of statements (currently 70), we use bsearch() now.
> >>>>
> >>>> Top of dump file:
> >>>> #v+
> >>>> --
> >>>> -- PostgreSQL database dump
> >>>> --
> >>>>
> >>>> SET client_encoding = 'UTF8';
> >>>> SET check_function_bodies = false;
> >>>> SET client_min_messages = warning;
> >>>>
> >>>> --
> >>>> -- Name: db_name; Type: DATABASE; Schema: -; Owner: user
> >>>> --
> >>>>
> >>>> CREATE DATABASE db_name WITH TEMPLATE = template0 ENCODING =  
> >>>> 'UTF8';
> >>>>
> >>>>
> >>>> ALTER DATABASE db_name OWNER TO user;
> >>>>
> >>>> \connect db_name
> >>>> #v-
> >>>>
> >>>> With pgpool-II 2.2 beta2 when I try create db from that dump got  
> >>>> such
> >>>> error in postgres log:
> >>>>
> >>>> #v+
> >>>> ERROR:  syntax error at or near "send" at character 1
> >>>> STATEMENT:  send invalid query from pgpool to abort transaction
> >>>> #v-
> >>>>
> >>>> With pgpool-II 2.2 beta1 everything works fine.
> >>>>
> >>>> pgpool work as:
> >>>> replication_mode = true
> >>>> replication_strict = true
> >>>> load_balance_mode = true
> >>>> replication_stop_on_mismatch = true
> >>>> replicate_select = false
> >>>> master_slave_mode = false
> >>>>
> >>>> Any idea why that happen ?
> >>>
> >>> Oops. CREATE DATABASE does not seem to be transaction safe. Please
> >>> apply included patches.
> >>> --
> >>> Tatsuo Ishii
> >>> SRA OSS, Inc. Japan
> >>> Index: pool_process_query.c
> >>> ===================================================================
> >>> RCS file: /cvsroot/pgpool/pgpool-II/pool_process_query.c,v
> >>> retrieving revision 1.137
> >>> diff -c -r1.137 pool_process_query.c
> >>> *** pool_process_query.c    6 Feb 2009 15:30:51 -0000    1.137
> >>> --- pool_process_query.c    9 Feb 2009 10:20:06 -0000
> >>> ***************
> >>> *** 3910,3916 ****
> >>>        T_ViewStmt,        /* CREATE VIEW */
> >>>        T_LoadStmt,
> >>>        T_CreateDomainStmt,
> >>> !        T_CreatedbStmt,
> >>>        T_DropdbStmt,
> >>>        T_CreateSeqStmt,
> >>>        T_AlterSeqStmt,
> >>> --- 3910,3918 ----
> >>>        T_ViewStmt,        /* CREATE VIEW */
> >>>        T_LoadStmt,
> >>>        T_CreateDomainStmt,
> >>> !        /*
> >>> !          T_CreatedbStmt,    CREATE DATABASE cannot execute  
> >>> inside a
> >>> transaction block
> >>> !        */
> >>>        T_DropdbStmt,
> >>>        T_CreateSeqStmt,
> >>>        T_AlterSeqStmt,
> >>> _______________________________________________
> >>> Pgpool-general mailing list
> >>> Pgpool-general at pgfoundry.org
> >>> http://pgfoundry.org/mailman/listinfo/pgpool-general
> >>