674655c641
Trying out to run processes via mpirun in Podman containers has shown
that the CMA btl_vader_single_copy_mechanism does not work when user
namespaces are involved.
Creating containers with Podman requires at least user namespaces to be
able to do unprivileged mounts in a container
Even if running the container with user namespace user ID mappings which
result in the same user ID on the inside and outside of all involved
containers, the check in the kernel to allow ptrace (and thus
process_vm_{read,write}v()), fails if the same IDs are not in the same
user namespace.
One workaround is to specify '--mca btl_vader_single_copy_mechanism none'
and this commit adds code to automatically skip CMA if user namespaces
are detected and fall back to MCA_BTL_VADER_EMUL.
Signed-off-by: Adrian Reber <areber@redhat.com>
(cherry picked from commit fc68d8a90fe86284e9dc730f878b55c0412f01d2)
157 строки
5.5 KiB
Plaintext
157 строки
5.5 KiB
Plaintext
# -*- text -*-
|
|
#
|
|
# Copyright (c) 2004-2009 The University of Tennessee and The University
|
|
# of Tennessee Research Foundation. All rights
|
|
# reserved.
|
|
# Copyright (c) 2006-2014 Cisco Systems, Inc. All rights reserved.
|
|
# Copyright (c) 2012-2014 Los Alamos National Security, LLC.
|
|
# All rights reserved.
|
|
# Copyright (c) 2014 Research Organization for Information Science
|
|
# and Technology (RIST). All rights reserved.
|
|
# $COPYRIGHT$
|
|
#
|
|
# Additional copyrights may follow
|
|
#
|
|
# $HEADER$
|
|
#
|
|
# This is the US/English help file for Open MPI's shared memory support.
|
|
#
|
|
[sys call fail]
|
|
A system call failed during vader shared memory BTL initialization
|
|
that should not have. It is likely that your MPI job will now either
|
|
abort or experience performance degradation.
|
|
|
|
System call: %s
|
|
Error: %s (errno %d)
|
|
#
|
|
[no locality]
|
|
WARNING: Missing locality information required for vader shared memory
|
|
BTL initialization. Continuing without shared memory support.
|
|
#
|
|
[knem permission denied]
|
|
WARING: Open MPI failed to open the /dev/knem device due to a
|
|
permissions problem. Please check with your system administrator to
|
|
get the permissions fixed, or set the btl_vader_single_copy_mechanism
|
|
MCA variable to none to silence this warning and run without knem
|
|
support.
|
|
|
|
Local host: %s
|
|
/dev/knem permissions: 0%o
|
|
#
|
|
[knem fail open]
|
|
WARNING: Open MPI failed to open the /dev/knem device due to a local
|
|
error. Please check with your system administrator to get the problem
|
|
fixed, or set the btl_vader_single_copy_mechanism MCA variable to none
|
|
to silence this warning and run without knem support.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
Errno: %d (%s)
|
|
#
|
|
[knem get ABI fail]
|
|
WARNING: Open MPI failed to retrieve the ABI version from the
|
|
/dev/knem device due to a local error. This usually indicates an
|
|
error in your knem installation; please check with your system
|
|
administrator, or set the btl_vader_single_copy_mechanism MCA variable
|
|
to none to silence this warning and run without knem support.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
Errno: %d (%s)
|
|
#
|
|
[knem ABI mismatch]
|
|
WARNING: Open MPI was compiled with support for one version of the
|
|
knem kernel module, but it discovered a different version running in
|
|
/dev/knem. Open MPI needs to be installed with support for the same
|
|
version of knem as is in the running Linux kernel. Please check with
|
|
your system administrator, or set the btl_vader_single_copy_mechanism
|
|
MCA variable to none to silence this warning and run without knem
|
|
support.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
Open MPI's knem version: 0x%x
|
|
/dev/knem's version: 0x%x
|
|
#
|
|
[knem mmap fail]
|
|
Open MPI failed to map support from the knem Linux kernel module; this
|
|
shouldn't happen. Please check with your system administrator, or set
|
|
the btl_vader_single_copy_mechanism MCA variable to none to silence
|
|
this warning and run without knem support.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
System call: mmap()
|
|
Errno: %d (%s)
|
|
#
|
|
[knem init error]
|
|
Open MPI encountered an error during the knem initialization. Please
|
|
check with your system administrator, or set the
|
|
btl_vader_single_copy_mechanism MCA variable to none to silence this
|
|
warning and run without knem support.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
System call: %s
|
|
Errno: %d (%s)
|
|
#
|
|
[knem requested but not available]
|
|
WARNING: Linux kernel knem support was requested via the
|
|
btl_vader_single_copy_mechanism MCA parameter, but Knem support was either not
|
|
compiled into this Open MPI installation, or Knem support was unable
|
|
to be activated in this process.
|
|
|
|
The vader BTL will fall back on another single-copy mechanism if one
|
|
is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
#
|
|
[cma-permission-denied]
|
|
WARNING: Linux kernel CMA support was requested via the
|
|
btl_vader_single_copy_mechanism MCA variable, but CMA support is
|
|
not available due to restrictive ptrace settings.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
#
|
|
[cma-different-user-namespace-error]
|
|
ERROR: Linux kernel CMA support was requested via the
|
|
btl_vader_single_copy_mechanism MCA variable, but CMA support is
|
|
not available due to different user namespaces.
|
|
|
|
Your MPI job will abort now. Please select another value for
|
|
btl_vader_single_copy_mechanism.
|
|
|
|
Local host: %s
|
|
#
|
|
[cma-different-user-namespace-warning]
|
|
WARNING: The default btl_vader_single_copy_mechanism CMA is
|
|
not available due to different user namespaces.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
#
|
|
[xpmem-make-failed]
|
|
WARNING: Could not generate an xpmem segment id for this process'
|
|
address space.
|
|
|
|
The vader shared memory BTL will fall back on another single-copy
|
|
mechanism if one is available. This may result in lower performance.
|
|
|
|
Local host: %s
|
|
Error code: %d (%s)
|