build report

Wei-keng Liao wkliao at eecs.northwestern.edu
Mon Sep 24 19:08:50 CDT 2018


The failed programs appear inconsistent when running different numbers of processes.
The error message reads to me likely an internal error of MPICH.
Assertion failed in file ../src/mpid/ch3/channels/nemesis/src/ch3_progress.c at line 530: payload_len >= sizeof (MPIDI_CH3_Pkt_t)

One of the failed program is examples/F90/get_info.f90 which only calls file
create, get MPI info object, and file close. It does not perform complex I/O
at all. I am wondering if this is caused by “make ptests” which runs multiple
MPI jobs immediately one after another. On some systems, users are suggested
to add a sleep command between 2 jobs.

Rob, have you seen this error?

Wei-keng

> On Sep 24, 2018, at 3:32 PM, Jim Edwards <jedwards at ucar.edu> wrote:
> 
> Building on an arm system (thunderx4) with mpich 3.2,  arm 18.4 compilers and a lustre filesystem.  I am getting errors from ptests, output attached.
> 
>  
> 
> -- 
> Jim Edwards
> 
> CESM Software Engineer
> National Center for Atmospheric Research
> Boulder, CO 
> <ptest.out>



More information about the parallel-netcdf mailing list