ESyS-Particle

some weird messages

Asked by Andrea Cortis

Hello Dion,

As you can see I have managed to successfully migrate to Launchpad.

Got a fresh Kubuntu 8.04 installation on my AMD 64bit machine with two-dual cores , followed your detailed instructions and run the GravityTut.py example, but I am not sure what all these Fatal messages mean ... ?

Anything to do with the fact that I am running Esys on a workstation and not on a cluster?

Many thanks!

ciao

Andrea

acortis@acortis-desktop:~$ mpirun -np 5 `which mpipython` GravityTut.py
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,1]: OpenIB on host acortis-desktop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,2]: OpenIB on host acortis-desktop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,4]: OpenIB on host acortis-desktop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,0]: OpenIB on host acortis-desktop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,3]: OpenIB on host acortis-desktop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
CSubLatticeControler::initMPI()
CSubLatticeControler::initMPI()
CSubLatticeControler::initMPI()
CSubLatticeControler::initMPI()
slave started at local/global rank 4230256 / 1
slave started at local/global rank 4230256 / 2
slave started at local/global rank 4230256 / 3
slave started at local/global rank 4230256 / 4

Question information

Language:
English Edit question
Status:
Answered
For:
ESyS-Particle Edit question
Assignee:
No assignee Edit question
Last query:
Last reply:
Revision history for this message
Dion Weatherley (d-weatherley) said : 		#1

Hi Andrea,

I get similar error messages on my laptop installation. It is related to OpenMPI but does not appear to be fatal for ESyS-Particle simulations. They run to completion properly. I'm tending to ignore it for now. I suspect that when OpenMPI gets updated in a future release these messages will go away.

Cheers,

Dion.

Revision history for this message
SteffenAbe (s-abe) said : 		#2

Definitely an OpenMPI problem. According to the error message OpenMPI expects to find an Infiniband network card (HCA). This most probably means that OpenMPI has been compiled with the wrong options, possible with Infiniband (OpenIB) as the only network driver. Could you send me the output of ompi_info on your machine ?

Steffen

Revision history for this message
Feng Chen (fchen3-gmail) said : 		#3

I met this problem too, and it is an OpenMPI problem which appears only on my laptop.

fchen@fchen-laptop:~/esp2/ex$ mpiexec -machinefile hosts.txt -np 2 mpipython bingle_output.py >simdata.csv
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,1]: OpenIB on host fchen-laptop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------
libibverbs: Fatal: couldn't read uverbs ABI version.
--------------------------------------------------------------------------
[0,1,0]: OpenIB on host fchen-laptop was unable to find any HCAs.
Another transport will be used instead, although this may result in
lower performance.
--------------------------------------------------------------------------

Although this does not affect the ESys-Particle solution, it is still quite annoying, I found a possible solution from the two posts:
http://www.open-mpi.org/community/lists/users/2007/09/4052.php
http://www.open-mpi.org/faq/?category=tuning#setting-mca-params:

By default, two files are searched (in order):
   1. $HOME/.openmpi/mca-params.conf: The user-supplied set of values takes the highest precedence.
   2. $prefix/etc/openmpi-mca-params.conf: The system-supplied set of values has a lower precedence.
--------------------------------------------------------------------------------------------------------------------------------------
And therefore the solution is: create the mca-params.conf file in:
$HOME/.openmpi/mca-params.conf
put a single line: btl=^openib
--------------------------------------------------------------------------------------------------------------------------------------
This should be able to suppress the OpenMPI error message.

Feng Chen

Revision history for this message
Dion Weatherley (d-weatherley) said : 		#4

Hi Feng Chen,

Thanks for the information. This is very helpful because, as you say, the error messages are very annoying.

Cheers,

Dion

Can you help with this problem?

Provide an answer of your own, or ask Andrea Cortis for more information if necessary.

To post a message you must log in.