Re: [Postgres-xc-general] How to use it?

[Koichi S. <koi...@gm...>]

Thank you very much for your interest in Postgres-XC.    I hope my
comments inline is helpful for you.
----------
Koichi Suzuki



2010/8/13 Dika Ye <di...@nt...>:
> Hi list,
>
>
>
> I am newer of postgres-xc, now, I am planning to create the postgresql
> cluster using this software, and here I have some questions:
>
>
>
> 1)       Can postgres-xc data node and coordinator install in the same
> machine, same working directory? How to deploy them?

A coordinator and  a data node can be installed in the same machine,
but they can't share the working directory.  From PostgreSQL point of
view, they're separete databases.    Please refer to "Install manual"
for practice.   Please write to me if you have further questions.

>
> 2)       If I set the coordinator use port 2000 and data node use port
> 12000, and which my app server will connect to?

App server should connect to 2000.

>
> 3)       Can I install more than one GTM server? And can they work on active
> / standby or active / active mode? How to configure it?

No.  We're now designing GTM standby and trying to assign a resource
for the implementation.

>
> 4)       The document descript how to start GTM, coordinator and data node,
> but don’t descript how to stop it, recovery it. When one of my data node
> failed, how to recovery it? And how the coordinator, GTM server……

gtm_ctl and pg_ctl utilities can stop these features, like PostgreSQL.
Michael: Could you provide the information how to stop them?

>
> 5)       I saw the architecture of postgre-xc, my understanding is that my
> app server will connect to coordinator server, then read / write into it,
> right?

Yes.   Data node is not directly visible from applications.

>
> 6)       Does it need to deploy a SAN/NAS/NFS file system for the data node
> servers or coordinator servers? Or the data keep a copy in each data node
> server or coordinator server?

No.   Each data node has its own database.    XC provides two ways of
table distribution, distributed and replicated.  In distributed table,
each tuple is distributed according to the value of distribution
column.   In replicated tables, tuples are replicated to all the data
nodes.

For the criteria which table should be distributed or replicated,
please take a look at my presentation material for PGCon2010 available
from Postgres-XC page in sourceforge.net.

>
> 7)       How many data node server or coordinator server does it support?
> How to add a new data node server or coordinator server into the current
> postgres-xc system?

So far, we tested up to ten coordinators and ten data nodes.    We
have to add coordinator or data node manually and it is not simple so
far.   We're planning to provide a utility for this by the end of this
calendar year.

>
>
>
> Is there any configuration sample? Where can find it?

Please take a look at installation manual and DBT-1 installation
document.   You will find an example.

Please feel free to write any questions, comments and requirements to
this mailing list.

Best Regards;
---
Koichi Suzuki

>
>
>
> Thanks.
>
>
>
> Best wishes,
>
>
>
> Dika.Ye
>
>
>
> ------------------------------------------------------------------------------
> This SF.net email is sponsored by
>
> Make an app they can't live without
> Enter the BlackBerry Developer Challenge
> http://p.sf.net/sfu/RIM-dev2dev
> _______________________________________________
> Postgres-xc-general mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-general
>
>