OpenMPI broken on host of virtual machine

[Kevin H. Hobbs]

Our F13 cluster is primarily used to run OpenMPI applications.

I installed a virtual guest using the instructions at :

http://docs.fedoraproject.org/en-US/Fedora/13/html/Virtualization_Guide/part-Virtualization-Installation.html

to do some development builds of a package I use and now OpenMPI jobs
fail to start with :

[bubbles.hooperlab][[55228,1],0][btl_tcp_endpoint.c:638:mca_btl_tcp_endpoint_complete_connect]
connect() to 192.168.122.1 failed: Connection refused (111)

on node 0, bubbles

192.168.122.1 belongs to virbr0 on bubbles

bubbles and all of the compute nodes are on 192.168.0.*

If I take down virbr0 on bubbles everything works.

Where might things be going wrong?

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 253 bytes
Desc: OpenPGP digital signature
Url : http://lists.fedoraproject.org/pipermail/users/attachments/20110505/23c6d228/attachment.bin