4. Slurm

Background

Resource schedulers are a pretty important piece of a supercomputer. Slurm is an open source one made by LLNL and is widely used. Almost all centers these days use slurm. It allows you to allocate resources, run programs either as a batch job or interactive.

Ensure uniform munged keys

cd /
tar -xzvf /opt/munge.tar.gz
sha256sum /etc/munge/munge.key


systemctl start munge.service
systemctl enable munge.service


cp /etc/slurm-llnl/cgroup.conf.example  /etc/slurm-llnl/cgroup.conf

# compute nodes
systemctl enable slurmd.service


systemctl enable slurmctld.service

refresh state

scontrol update nodename=mini[1-5] state=resume scontrol update nodename=mini[1-5] state=unknown

Lmod

Follow instructions here: https://lmod.readthedocs.io/en/latest/030_installing.html

Lua

wget https://sourceforge.net/projects/lmod/files/lua-5.1.4.9.tar.bz2
./configure --prefix=/opt/cc/lua/5.1.4.9
# Add #include <string.h> to the unix package that breaks
make -j6 && sudo make install
ln -s /opt/cc/lua/lua/bin/lua /usr/local/bin
ln -s /opt/cc/lua/lua/bin/luac /usr/local/bin

Lmod

sudo pacman -Sy tcl
wget https://sourceforge.net/projects/lmod/files/Lmod-8.7.tar.bz2/download
tar xf download
cd Lmod-8.7
# Install lmod to /opt/cc/lmod/8.7 and modulefiles to /opt/modulefiles
./configure --prefix=/opt/cc --with-module-root-path=/opt/modulefiles
sudo ln -s /opt/cc/lmod/lmod/init/profile        /etc/profile.d/z00_lmod.sh
sudo ln -s /opt/cc/lmod/lmod/init/cshrc          /etc/profile.d/z00_lmod.csh
# # no fish ln # shell -s /opt/cc/lmod/lmod/init/profile.fish   /etc/fish/conf.d/z00_lmod.fish

Testing out

type module
echo $LMOD_CMD
echo $MODULEPATH

Compute nodes

salloc -N5
srun pacman -Sy tcl --noconfirm
srun ln -s /opt/cc/lua/lua/bin/lua /usr/local/bin
srun ln -s /opt/cc/lua/lua/bin/luac /usr/local/bin
srun ln -s /opt/cc/lmod/lmod/init/profile        /etc/profile.d/z00_lmod.sh
srun ln -s /opt/cc/lmod/lmod/init/cshrc          /etc/profile.d/z00_lmod.csh