Re: [Starcluster] cluster start-up hangs when starting Sun Grid Engine

From: Damian Eads <no email>
Date: Sat, 17 Apr 2010 15:31:50 -0400

On Sat, Apr 17, 2010 at 11:09 AM, Thomas Deselaers
<deselaers_at_vision.ee.ethz.ch> wrote:
> On Fri, Apr 16, 2010 at 21:01, Justin Riley <jtriley_at_mit.edu> wrote:
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> Hi Damian,
>>
>> You're experiencing the same problem as Gabriel earlier. Basically with
>> the 64bit StarCluster AMI it has a problem with NFS and this is what
>> you're experiencing.
>
> Hi,
>
> Justin, just to let you know: I have also been using the 64bit amis
> and have not had any problems at all.
>
> Cheers,
> thomas

Hi Thomas,

I don't think the issue was 64-bit AMIs but 64-bit AMIs on certain
machines, and in particular, 8 core machines like c1.xlarge and
m2.xlarge. I could sometimes get the AMI to start on 8 core machines
(maybe because my configuration incorrectly specified m1.large?) but
most of the time it would hang at cluster creation time. You may be
able to use all 8 cores manually by spawning processes at the shell,
ssh, fork(), or pthreads but can you use MPI, grid, or any other job
management tool on 8 core machines using the older 64-bit AMI. Please
clarify how you're using the older 64-bit AMIs and on what
architecture?

Thanks,

Damian
Received on Sat Apr 17 2010 - 15:31:51 EDT