Re: [Postgres-xc-general] multiple gtm_proxy configuration

[Sandeep G. <gup...@gm...>]

Hi,

 I digged a little deeper for this case. Something is amiss and was
wondering if I could get some help. When I launch
multiple clients simultaneously, where each client is performing a copy on
the same table.

First, thing that looks off is that the gtm_proxy has too high activity.
ps command show 428% cpu and 17 % memory on 64 GB ram machine.

6795 sandeep   20   0 8468m 8.1g  928 S  428 17.1  39:00.17 gtm_proxy

This memory usage jumps to 40%. This I think is very high.

Second, the individual postgres processes (both coordinator and datanodes)
do no show cpu activity. Their cpu usage is 0.

Third, I get a warning which is fine. However, I feel it makes the system
unstable
WARNING:  worker took too long to start; canceled
WARNING:  worker took too long to start; canceled
FATAL:  Can not register Datanode on GTM


In summary, there is something amiss. I would have expected copy command to
work fine because looking at the code it doesn't to  be heavily using the
GTM.  Any fixes or ideas would be greatly welcomed.


-Sandeep






On Mon, Oct 7, 2013 at 11:51 AM, Sandeep Gupta <gup...@gm...>wrote:

> Hi,
>
>  I have a short query. In my setup I have one gtm, one gtm_proxy, one
> co-ordinator, and, multiple datanodes.
>
> It seems for copy gtm_proxy seems to be a bottleneck. Is there any way I
> can have multiple gtm_proxies. If so, how can I use in my setup. Just point
> me to the documentation.
>
> Thanks.
> Sandeep
>
>