postgres-xc-developers Mailing List for Postgres-XC

Another info:

Transactions involved by the failed datanode will fail.     When a
coordinator detects failure of a datanode, it aborts live transactions
the failed datanode is involved.   When multiple datanodes are
involved, coordinator will abort the transaction at all such
datanodes.

After then, transactions involved by the failed node will fail too.

When the crashed datanode is recovered and rejoins the cluster,
transactions running at the time of crash may remain "prepared",
"committed" or "aborted".   We need to cleanup them and synchronize
with other cluster node status.   I'm writing a tool "pgxc_clean" to
take care of this (so far, we have to do some tricky step to clean
this up) and I do hope this is in V1.0.

I'm afraid this explanation is not simple.

Best Regards;
----------
Koichi Suzuki



2012年3月20日19:41 Ashutosh Bapat <ash...@en...>:
> Hi Kris,
> About choosing node to read in replicated tables - Every coordinator has a
> list of preferred nodes. If a table is replicated on a preferred node, that
> node is chosen to read.
>
>
> On Tue, Mar 20, 2012 at 3:49 PM, Krzysztof Nowicki
> <krz...@gm...> wrote:
>>
>> Hi all,
>>
>> I can't find information about some aspects of postgres-xc, maybe
>> someone could quick answer to the following questions:
>>
>> Where GTM stores data ?
>>
>> Which algorithm is used to choose node for READ in replicated tables ?
>>
>> When some of data node crash and then after recover if data will
>> automatically be synchronize ?
>>
>> What happened with transaction when during processing one of the node
>> in the cluster crash ( rollback or commited) ? What happened with next
>> transactions ?
>>
>> BR,
>>
>> Kris Nowicki
>>
>>
>> ------------------------------------------------------------------------------
>> This SF email is sponsosred by:
>> Try Windows Azure free for 90 days Click Here
>> http://p.sf.net/sfu/sfd2d-msazure
>> _______________________________________________
>> Postgres-xc-developers mailing list
>> Pos...@li...
>> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>
>
>
>
> --
> Best Wishes,
> Ashutosh Bapat
> EntepriseDB Corporation
> The Enterprise Postgres Company
>
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> Postgres-xc-developers mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>

Hi Kris,
About choosing node to read in replicated tables - Every coordinator has a
list of preferred nodes. If a table is replicated on a preferred node, that
node is chosen to read.

On Tue, Mar 20, 2012 at 3:49 PM, Krzysztof Nowicki <
krz...@gm...> wrote:

> Hi all,
>
> I can't find information about some aspects of postgres-xc, maybe
> someone could quick answer to the following questions:
>
> Where GTM stores data ?
>
> Which algorithm is used to choose node for READ in replicated tables ?
>
> When some of data node crash and then after recover if data will
> automatically be synchronize ?
>
> What happened with transaction when during processing one of the node
> in the cluster crash ( rollback or commited) ? What happened with next
> transactions ?
>
> BR,
>
> Kris Nowicki
>
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
> _______________________________________________
> Postgres-xc-developers mailing list
> Pos...@li...
> https://lists.sourceforge.net/lists/listinfo/postgres-xc-developers
>



-- 
Best Wishes,
Ashutosh Bapat
EntepriseDB Corporation
The Enterprise Postgres Company

Hi all,

I can't find information about some aspects of postgres-xc, maybe
someone could quick answer to the following questions:

Where GTM stores data ?

Which algorithm is used to choose node for READ in replicated tables ?

When some of data node crash and then after recover if data will
automatically be synchronize ?

What happened with transaction when during processing one of the node
in the cluster crash ( rollback or commited) ? What happened with next
transactions ?

BR,

Kris Nowicki