[torqueusers] Problem with hwloc

[Jon Tegner]

Hi!

I have issues with hwloc on a set of nodes. When starting pbs_mom on the
nodes I get the error message

* hwloc 1.11.2 has encountered what looks like an error from the
operating system.
*
* L3 (cpuset 0x00000003,0xf0000000) intersects with NUMANode (P#2 cpuset
0x3f000000) without inclusion!
* Error occurred in topology.c line 1046

The nodes were previously working without problems using CentOS-6.5, and
the problem occurred when switching to CentOS-7.3.

I have previously, on a different set of hardware, since similar
problems (also on CentOS-7.3), but in that case I could resolve the
issues by downgrade torque, torque-libs, torque-client, torque-mom and
torque-devel from 4.2.10-10 to 4.2.10-5.

I'm a bit confused since (as I wrote above):

* The nodes in question were working without problems on CentOS-6.5.

* Different hardware is working without problem on the same OS using the
same version of torque (4.2.10-5).

Thanks!

/jon

[Chad Vizino]

Have you tried building Torque with a newer version of hwloc? The latest is
1.11.7 available here:

https://www.open-mpi.org/software/hwloc/v1.11/

-Chad

Post by Jon Tegner
Hi!
I have issues with hwloc on a set of nodes. When starting pbs_mom on the
nodes I get the error message
* hwloc 1.11.2 has encountered what looks like an error from the
operating system.
*
* L3 (cpuset 0x00000003,0xf0000000) intersects with NUMANode (P#2 cpuset
0x3f000000) without inclusion!
* Error occurred in topology.c line 1046
The nodes were previously working without problems using CentOS-6.5, and
the problem occurred when switching to CentOS-7.3.
I have previously, on a different set of hardware, since similar
problems (also on CentOS-7.3), but in that case I could resolve the
issues by downgrade torque, torque-libs, torque-client, torque-mom and
torque-devel from 4.2.10-10 to 4.2.10-5.
* The nodes in question were working without problems on CentOS-6.5.
* Different hardware is working without problem on the same OS using the
same version of torque (4.2.10-5).
Thanks!
/jon
_______________________________________________
torqueusers mailing list
http://www.supercluster.org/mailman/listinfo/torqueusers