[Postgres-xl-bugs] problems when porting postgres-xl 9.2 to ppc64 linux

Hui NJ Shen shenhnj at cn.ibm.com
Mon Sep 8 18:39:23 PDT 2014


many thanks for your suggestions Mason!

I have verified the simple cluster with 1 GTM ,1 coordinator ,1 datanode.
It works fine. I think it's just like postgreSQL in this simple scenario.

I have also verified on Little Endian linux (ubuntu) on power , It works
fine with 1 GTM ,1 coordinator,2 datanode. so it actually is related with
BE/LE.


Best Regards
Shen Hui (沈慧)
Channel Technical Sales Specialist

Mobile: 15195886872
E-mail: shenhnj at cn.ibm.com

南京市鼓楼区汉中路2号世界贸易中心
World Trade Center,No. 2 HanZhong Road, GuLou District, NanJing,China




From:	Mason Sharp <msharp at translattice.com>
To:	Hui NJ Shen/China/IBM at IBMCN
Cc:	Nikhil Sontakke <nikhils at translattice.com>,
            "postgres-xl-bugs at lists.sourceforge.net"
            <postgres-xl-bugs at lists.sourceforge.net>
Date:	2014/09/06 00:31
Subject:	Re: [Postgres-xl-bugs] problems when porting postgres-xl 9.2 to
            ppc64 linux






On Fri, Sep 5, 2014 at 2:12 AM, Hui NJ Shen <shenhnj at cn.ibm.com> wrote:
  hi,Nikhil:

    thanks!

   Was the cluster created from scratch using pgxc_ctl? //yes I created it
  using pgxc_ctl;

  Was any data pre-loaded on the cluster or to the datanodes using "execute
  direct"?  //no data pre-loaded

  Please run "select * from pgxc_node" on all nodes and ensure that there
  is only ONE and SAME entry for the coordinator on all of them. Also two
  datanode entries should be consistent and same on all nodes as well. Pay
  special attention to the "node_type" column entries. //yes,I checked it
  again.it's right.


  would you give a suggestion on next action for me ?



A couple of thoughts-

Try setting log_statement = all in postgresql.conf on all of the nodes and
see if the command is being reissued somehow (a datanode thinking it is a
coordinator, for example).

Also, try a simpler cluster- 1 GTM, 1 Coordinator, and 1 Datanode.  If this
fails, maybe there is some PPC specific issue for some reason.

From there try 2 Datanodes, then 2 Coordinators, etc.

Thanks,

Mason



  Best Regards
  Shen Hui (沈慧)
  Channel Technical Sales Specialist

  Mobile: 15195886872
  E-mail: shenhnj at cn.ibm.com

  南京市鼓楼区汉中路2号世界贸易中心
  World Trade Center,No. 2 HanZhong Road, GuLou District, NanJing,China


  Inactive hide details for Nikhil Sontakke ---2014/09/05 14:02:17---Hi
  Hui, This is indeed the first time that we are seeing sucNikhil Sontakke
  ---2014/09/05 14:02:17---Hi Hui, This is indeed the first time that we
  are seeing such an error. The error

  From: Nikhil Sontakke <nikhils at translattice.com>
  To: Hui NJ Shen/China/IBM at IBMCN
  Cc: Mason Sharp <msharp at translattice.com>, "
  postgres-xl-bugs at lists.sourceforge.net" <
  postgres-xl-bugs at lists.sourceforge.net>
  Date: 2014/09/05 14:02



  Subject: Re: [Postgres-xl-bugs] problems when porting postgres-xl 9.2 to
  ppc64 linux



  Hi Hui,

  This is indeed the first time that we are seeing such an error. The error
  seems to be coming from within the pg_catalog operations where the
  corresponding composite type for the table "test" is being added as part
  of "create table test" command. That is basically existing PG code that
  XL has leveraged and that should run ok on little/big endian machines. So
  on the face of it, does not appear to be a distributed command being
  re-issued issue.. However, it's very difficult to say.

  Again to be clear:

  Was the cluster created from scratch using pgxc_ctl?

  Was any data pre-loaded on the cluster or to the datanodes using "execute
  direct"?

  Please run "select * from pgxc_node" on all nodes and ensure that there
  is only ONE and SAME entry for the coordinator on all of them. Also two
  datanode entries should be consistent and same on all nodes as well. Pay
  special attention to the "node_type" column entries.

  Hope this helps,
  Regards,
  NIkhils

  On Fri, Sep 5, 2014 at 9:32 AM, Hui NJ Shen <shenhnj at cn.ibm.com> wrote:
        hi,Mason:

           is it caused by a configuration error ?

        Best Regards
        Shen Hui (沈慧)
        Channel Technical Sales Specialist

        Mobile: 15195886872
        E-mail: shenhnj at cn.ibm.com

        南京市鼓楼区汉中路2号世界贸易中心
        World Trade Center,No. 2 HanZhong Road, GuLou District,
        NanJing,China


        Inactive hide details for Mason Sharp ---2014/09/04 03:04:33---On
        Wed, Sep 3, 2014 at 2:36 PM, Hui NJ Shen <shenhnj at cn.ibm.com>Mason
        Sharp ---2014/09/04 03:04:33---On Wed, Sep 3, 2014 at 2:36 PM, Hui
        NJ Shen <shenhnj at cn.ibm.com> wrote: > hi,mason:

        From: Mason Sharp <msharp at translattice.com>
        To: Hui NJ Shen/China/IBM at IBMCN
        Cc: Nikhil Sontakke <nikhils at translattice.com>, "
        postgres-xl-bugs at lists.sourceforge.net" <
        postgres-xl-bugs at lists.sourceforge.net>
        Date: 2014/09/04 03:04
        Subject: Re: [Postgres-xl-bugs] problems when porting postgres-xl
        9.2 to ppc64 linux




        On Wed, Sep 3, 2014 at 2:36 PM, Hui NJ Shen <shenhnj at cn.ibm.com>
        wrote:
              hi,mason:

                finally with your help ,I create the cluster with
              "pgxc_ctl" ,but seems the same error occur.

                postgres=# create table test(sh int);
              ERROR:  duplicate key value violates unique constraint
              "pg_type_typname_nsp_index"
              DETAIL:  Key (typname, typnamespace)=(test, 2200) already
              exists.
              postgres=#

        Can you please email your pgxc_ctl.conf file?

        It is almost as if the coordinator is trying to contact the same
        datanode twice. Or, a datanode thinks it is a coordinator and is
        trying to contact other datanodes. pgxc_ctl normally catches host
        and port conflicts, however.

        Also, are you using the latest code here (although it should not
        matter):

        git clone git://git.code.sf.net/p/postgres-xl/postgres-xl
        postgres-xl-postgres-xl

        Can you run createdb and create a new database and use that?

        Thanks,

        Mason






  --
  TransLattice - http://www.translattice.com
  Distributed and Clustered Database Solutions
  Postgres-XL Support






--
Mason Sharp

TransLattice - http://www.translattice.com
TransLattice Elastic Database (TED)
Postgres-XL Support
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.postgres-xl.org/private.cgi/postgres-xl-bugs-postgres-xl.org/attachments/20140909/c0e0f70d/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: graycol.gif
Type: image/gif
Size: 105 bytes
Desc: not available
URL: <http://lists.postgres-xl.org/private.cgi/postgres-xl-bugs-postgres-xl.org/attachments/20140909/c0e0f70d/attachment.gif>


More information about the postgres-xl-bugs mailing list