I have read article on https://podman.io/blogs/2019/09/26/podman-in-hpc.html. And I have built openmpi image with fedroa.30 and centos.7.9.2009. The hello_c runs successfully on fedroa.30 image. But hello_c failed to run on centos.7.9.2009. It hang when runs on centos.7.9.2009 image. I run it with -d, and it hang at "[rsyncdk2:2674797] [[58966,0],0] Releasing job data for [INVALID]". The following is the whole message. [rsync@rsyncdk2 ~] mpirun -d -v -host rsyncdk2:2 podman run --rm --env-host --userns=keep-id --network=host --pid=host --ipc=host 69fb54f4bb75 /home/hello_c [rsyncdk2:2674797] procdir: /tmp/ompi.rsyncdk2.1000/pid.2674797/0/0 [rsyncdk2:2674797] jobdir: /tmp/ompi.rsyncdk2.1000/pid.2674797/0 [rsyncdk2:2674797] top: /tmp/ompi.rsyncdk2.1000/pid.2674797 [rsyncdk2:2674797] top: /tmp/ompi.rsyncdk2.1000 [rsyncdk2:2674797] tmp: /tmp [rsyncdk2:2674797] sess_dir_cleanup: job session dir does not exist [rsyncdk2:2674797] sess_dir_cleanup: top session dir not empty - leaving [rsyncdk2:2674797] procdir: /tmp/ompi.rsyncdk2.1000/pid.2674797/0/0 [rsyncdk2:2674797] jobdir: /tmp/ompi.rsyncdk2.1000/pid.2674797/0 [rsyncdk2:2674797] top: /tmp/ompi.rsyncdk2.1000/pid.2674797 [rsyncdk2:2674797] top: /tmp/ompi.rsyncdk2.1000 [rsyncdk2:2674797] tmp: /tmp [rsyncdk2:2674797] [[58966,0],0] Releasing job data for [INVALID]
On Fri, Feb 18, 2022 at 04:55:16AM -0000, None via Podman wrote:
I have read article on https://podman.io/blogs/2019/09/26/podman-in-hpc.html.
That is my article.
And I have built openmpi image with fedroa.30 and centos.7.9.2009. The hello_c runs successfully on fedroa.30 image. But hello_c failed to run on centos.7.9.2009. It hang when runs on centos.7.9.2009 image.
Which version of Open MPI do you have have in each container and on your host? Adrian
I'm glad to contact you. Host MPI: openmpi-4.1.1 Fedora 30 MPI: openmpi-3.1.4 Centos 7.9 MPI: openmpi-1.10.7 I have changed to openmpi-4.1.1 on centos 7.9. It runs successfully. Thanks for the tip. GHui
participants (2)
-
Adrian Reber -
ugiwgh@qq.com