Restart file problem

LAMMPS LAMMPS Beginners
,
1

I’m running a job in LAMMPS on an HPC system. The HPC center gives each user 3 days, after which we need to queue again to run the job. Therefore, I need to restart the file, but I’m getting the following error. My LAMMPS version is “lammps-stable_29Aug2024”. The HPC center uses Linux. I’m running this restart file on the 22 July 2025 Windows LAMMPS version. So the file isn’t faulty. In the HPC center, I started and continued with the same version of LAMMPS, so the error isn’t related to the version. Please help me, I don’t want my work to be ruined. If you want anything else, I can share it.

EROR:

SLURM_NODELIST orfoz262
NUMBER OF CORES 56
Loading modulefiles version 2021.11
Loading itac version 2022.0

Loading apps/lammps/29Aug2024_stable_oneapi-2024
Loading requirement: comp/oneapi/2024-modules/compiler-rt32/latest
comp/oneapi/2024-modules/ifort32/2024.0.0
comp/oneapi/2024-modules/tbb/latest
comp/oneapi/2024-modules/compiler-rt/latest
comp/oneapi/2024-modules/mkl/2024.0 comp/oneapi/2024-modules/ifort/2024.0.0
comp/oneapi/2024-modules/debugger/2024.0.0
comp/oneapi/2024-modules/dpct/2024.0.0 comp/oneapi/2024-modules/dal/2024.0.0
comp/oneapi/2024-modules/tbb32/latest
comp/oneapi/2024-modules/intel_ipp_ia32/2021.10
comp/oneapi/2024-modules/dnnl/3.3.0 comp/oneapi/2024-modules/tbb/2021.11
comp/oneapi/2024-modules/dev-utilities/2024.0.0
comp/oneapi/2024-modules/mkl32/2024.0 comp/oneapi/2024-modules/ccl/2021.11.0
comp/oneapi/2024-modules/intel_ippcp_ia32/2021.9
comp/oneapi/2024-modules/compiler32/2024.0.0
comp/oneapi/2024-modules/compiler-rt/2024.0.0
comp/oneapi/2024-modules/inspector/2024.0
comp/oneapi/2024-modules/oclfpga/2024.0.0
comp/oneapi/2024-modules/mpi/2021.11
comp/oneapi/2024-modules/compiler-rt32/2024.0.0
comp/oneapi/2024-modules/advisor/2024.0
comp/oneapi/2024-modules/intel_ipp_intel64/2021.10
comp/oneapi/2024-modules/intel_ippcp_intel64/2021.9
comp/oneapi/2024-modules/tbb32/2021.11 comp/oneapi/2024-modules/dpl/2022.3
comp/oneapi/2024-modules/compiler/2024.0.0
comp/oneapi/2024-modules/itac/2022.0 comp/oneapi/2024-modules/vtune/2024.0
comp/oneapi/2024
LAMMPS (29 Aug 2024)
using 1 OpenMP thread(s) per MPI task
Reading restart file …
restart file = 29 Aug 2024, LAMMPS = 29 Aug 2024
WARNING: Restart file used different # of processors: 112 vs. 56 (src/applications/lammps/lammps-stable_29Aug2024/src/read_restart.cpp:628)
restoring atom style full from restart
ERROR: Invalid flag in header section of restart file (src/applications/lammps/lammps-stable_29Aug2024/src/read_restart.cpp:855)
Last command: read_restart file.restart.4000000

2

@ihs Your logic is not 100% correct. However, it is extremely difficult to debug such matters from remote since part of the problem may be the executable.

Since you confirm restarting to work on Windows, albeit with a different LAMMPS version, there are two main possible causes:

  • there is a bug in the LAMMPS version on the Linux cluster that has since been fixed. I noticed that it is using the original 29Aug2024 stable version and not the latest bugfix update. In fact there has been a bug fixed in the update, but that is with a very rarely used feature

  • the Intel compiler used to compile LAMMPS has created a broken executable. That is - unfortunately - not uncommon for recent Intel compiler releases. It is rather random, but seems to be more common for .0 version releases (which appears to be used). we have stopped testing LAMMPS with Intel compilers for that reason. Outside of the INTEL package, there is nothing in LAMMPS that will not run as fast with the GCC or Clang compilers.

I have two recommendations that you can try:

3

Hello,
I also tried the restart file with another version installed on the HPC (I think it’s the latest version, “22Jul2025_update1-voronoi-oneapi-2024”). The simulation is currently continuing from where it left off. Thank you for your interest and help.