Charm++ parallel programming system

[Phil Miller <mille121 AT illinois.edu>]

Hi Hemanth,

On the verbs-*, netlrts-*, and net-* builds of the Charm RTS, charmrun
attempts to launch compute processes via SSH to the compute nodes.

If you're running on just a single node, you could use ++local to tell
it to just fork() processes within that node.

If you're running on multiple nodes, on a cluster with MPI configured
using a launcher such as mpirun, mpiexec, ibrun, srun, etc, have a
look at the documentation for options
* ++mpiexec
* ++mpiexec-no-n
* ++remote-shell
here: 
http://charm.cs.illinois.edu/manuals/html/charm++/C.html#SECTION05321000000000000000

If your cluster has no such configuration already in place, you'll
need to make arrangements for passwordless SSH between whatever node
you launch jobs from and your compute nodes.

Phil

On Tue, Jun 14, 2016 at 12:11 PM, Kolla, Hemanth NMN 
<hnkolla AT sandia.gov>
 wrote:
> Hi,
>
> I'm trying to get some basic profiling of a code but am facing some teething
> run-time issues when I compile with tracemode projections. I'm running on an
> infiniband linux cluster in interactive mode, and when I launch the run I
> get a message
>
> "FIPS integrity verification test failed."
>
> and then it asks me for a passcode again, as though its trying to log me in
> once again. This happens for a regular executable as well as a *.prj
> executable.
>
> I built charm++ --with-production and --enable-projections for a
> verbs-linux-x86_64 target, and that build itself was smooth. I'd be grateful
> for any help.
>
> Thanks,
> Hemanth Kolla.
>
>