Am 03.03.2014 um 23:54 schrieb Gus Correa:
On 03/03/2014 04:36 PM, Ron Palmer wrote:
Thanks Reuti for your comments. I will peruse that FAQ detail.
I just thought of the fact that each of these rack computers have 4 ethernet sockets, eth0 - eth3... I could connect the cluster on a separate ethernet sockets via an extra switch not connected to the internet or any other computers, and accept all communication among them, and keep iptables up on the ethx connected to the outside world. I guess I would have to set up routing tables or something. Ah, more reading :-)
Thanks for your help. Ron
Hi Ron
If those extra interfaces are not in use, and if you have a spare switch, you can setup a separate private subnet exclusively for MPI. You need to configure the interfaces consistently (IP, subnet mask, perhaps a gateway). Configuring them statically is easy:
https://access.redhat.com/site/documentation//en-US/Red_Hat_Enterprise_Linux...
Use a subnet that doesn't intersect the existent/original IP range.
http://en.wikipedia.org/wiki/Private_network
You could also create host names associated to those IPs (say node01, node02, node02), resolve them via /etc/hosts on each computer, set passwordless ssh across these newly named "hosts". This may simpler/safer than messing with the iptables.
[Actually, the IP addresses you showed 192.168.X.Y, sound as a private subnet already, not Internet, but that may be the subnet for your organization/school/department already. So, you may set up a different one on these three computers for MPI and very-local access.]
Yep, maybe in the 10.0.0.0/8 range. BTW: are you setting up a cluster from scratch - will you also add any queuing system later on?
OpenMPI allows you to choose the interface that it will use, so you can direct it to your very-local subnet:
MPICH: http://wiki.mpich.org/mpich/index.php/Using_the_Hydra_Process_Manager#Hydra_... It should work for other Ethernet interfaces too. Nevertheless: in any case it might be easier for the applications to use the primary interface solely for MPI, and any other one for the external access (this is usually my setup, as I don't have to provide the assignment of other interfaces this way). -- Reuti
I hope this helps, Gus
On 4/03/2014 00:13, Reuti wrote:
Am 03.03.2014 um 01:34 schrieb Ron Palmer:
Gus, I have just replied with the details of the success, but I will clarify your questions here, if it helps next time.
Re the actual application to be run, 'inversion', I have only received binaries. I used to run them on gainsborough (without mpi) and that worked fine.
Home directories are not nfs shared, they are individual and separate, only the name is repeated.
I had tried password free ssh in all directions and permutations.
I had iptables down on sargeant and up on the other two.
Yes, I installed the gcc-gfortran.x86_64 AFTER I took those screnshots, and the post install output was identical to the top one (sargeant).
I am unsure about cpi and fortran...
Stuff remaining to get sorted out: 1. Get that hyperthreading set up - what is your suggestions? Disable and let mpi manage the cores? It depends on your applications, whether you can make use of it. Test it with HT switched off, switched on and increase the number of processes, until you see that any process is no longer running at 100% and you judge that you don't tolerate this slow down.
It's not uncommon to see an improvement up to 150% with HT turned on, but not 200% (depending on the workload).
2. run mpiexec with iptables up, need to figure out what traffic to allow. https://wiki.mpich.org/mpich/index.php/Frequently_Asked_Questions#Q:_How_do_...
-- Reuti
Great many thanks to all, and Gus, Reuti and Rajeev in particular.
Cheers, Ron
On 3/03/2014 10:09, Gustavo Correa wrote:
Hi Ron
_______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss
_______________________________________________ discuss mailing list [email protected] To manage subscription options or unsubscribe: https://lists.mpich.org/mailman/listinfo/discuss