Re: [Starpu-devel] starpu / guix on plafrim partition dedicated for the November HPC school

[Nathalie Furmento <nathalie.furmento@labri.fr>]

Manu,

Could you please send me the environment i need to call guix ?

Here what i get.

[hpcs-furment@miriel001 ~]$ guix build starpu --no-grafts --check
guix build: erreur : starpu : paquet inconnu
[hpcs-furment@miriel001 ~]$

Thanks!

Nathalie

On 22/10/2019 21:25, Emmanuel Agullo wrote:

Dear StarPU team,

I succeed to build starpu with guix on:
- my laptop,
- plafrim frontal node (devel13),
- a computing node on plafrim of type miriel (miriel015).

To do that, on any of those machines I run:

guix build starpu --no-grafts --check

And it's all working great.

However, when I am trying to build it on the plafrim-hpcs
machine (the partition of plafrim that we'll use for the HPC school in
November), I obtain an error during the starpu check step. I get that
both on the frontal node of plafrim-hpcs (mistral01) and on
a compute node (miriel001).

The failure is related to starpu_machine_display:

FAIL: starpu_machine_display
============================
[starpu][check_bus_config_file] No performance model for the bus, calibrating...
[starpu][check_bus_config_file] ... done
[error] `./starpu_machine_display' killed with signal 11; test marked as failed
warning: core file may not match specified executable file.
[New LWP 24848]
[Thread debugging using libthread_db enabled]
Using host libthread_db library "/gnu/store/ahqgl4h89xqj695lgqvsaf6zh2nhy4pj-glibc-2.29/lib/libthread_db.so.1".
Core was generated by `/tmp/guix-build-starpu-1.3.2.drv-0/source/tools/.libs/starpu_machine_display'.
Program terminated with signal SIGSEGV, Segmentation fault.
#0  starpu_tree_get_neighbour (tree=tree@entry=0x41d400, node=node@entry=0x41d400, visited=visited@entry=0x7fffffffc5e8 "\001", present=present@entry=0x41d594 "\001") at core/tree.c:113
113		for(st = 0; st < father->arity; st++)

I'm attaching the log of the error (starting at the check step).

Here is my guix environment (the same on all 5 machines), if it may
help.

eagullo@tek9:/tmp/toto$ guix describe --format=channels
(list (channel
        (name 'guix-hpc-non-free)
        (url "https://gitlab.inria.fr/guix-hpc/guix-hpc-non-free.git")
        (commit
          "0822b7eebb63e2360125e017934ba817686d9669"))
      (channel
        (name 'guix-hpc)
        (url "https://gitlab.inria.fr/guix-hpc/guix-hpc.git")
        (commit
          "b24a8fa55bfd4a0987320e0e45d3a85f0920427c"))
      (channel
        (name 'guix)
        (url "https://git.savannah.gnu.org/git/guix.git")
        (commit
          "bfaa06171842225c8bf4c839e35ac36a0b2c2d59")))

In case you cannot reproduce the bug on one of your machines,
and you need to check it out more closely, we can do that with my
account on this particular plafrim partition if you want to.

Note that there is no emergency to track it if it's not an obvious
issue I am missing (I can simply retrieve the pre-built binaries
with "guix build starpu", i.e. without the "--no-grafts --check"
options).

Thanks much in advance for you help once again!

Best,

Manu

_______________________________________________
Starpu-devel mailing list
Starpu-devel@lists.gforge.inria.fr
https://lists.gforge.inria.fr/mailman/listinfo/starpu-devel

-- 
Dr Nathalie Furmento               Ingénieure de recherche CNRS
http://www.labri.fr/~furmento/     LaBRI - Université Bordeaux