diff --git a/create_lanes.sh b/create_lanes.sh index ff5724d..75439f0 100755 --- a/create_lanes.sh +++ b/create_lanes.sh @@ -1,212 +1,212 @@ #!/bin/bash set -euo pipefail # Script made from the page # https://docs.ycrc.yale.edu/clusters-at-yale/guides/cryosparc/ script_path=$(dirname "$0") # relative script_path=$(cd "${script_path}" && pwd) # absolutized and normalized install_path="$HOME"/cryosparc # set up some more paths db_path=${install_path}/database worker_path=${install_path}/cryosparc2_worker # Usage usage () { echo "Usage:" echo " -p install path : prefix for installation [${install_path}] " echo " -v : be verbose" echo " -h : print this notice" echo "" } VERBOSE=false # Parse options while getopts ":p:vh" opt; do case $opt in p) install_path="${OPTARG}" ;; v) VERBOSE=true ;; h) usage OPTIND=1 exit 0 ;; \?) echo "Invalid option: -$OPTARG" >&2 usage OPTIND=1 exit 1 ;; :) echo "Option -$OPTARG requires an argument." >&2 usage OPTIND=1 exit 1 ;; esac done # Reset OPTIND to allow the next invocation to work OPTIND=1 message() { if $VERBOSE then echo "${1}" fi } echo "[Info] Starting the master if needed" if [ $(${install_path}/cryosparc_master/bin/cryosparcm.sh status | grep -c "CryoSPARC is not running") -eq 1 ]; then ${install_path}/cryosparc_master/bin/cryosparcm.sh start fi user_name=$(whoami) user_accounts=$(sacctmgr show assoc where user=${user_name} format=Account%100 -P | grep -v Account) mkdir -p ${install_path}/site_configs && cd ${install_path}/site_configs for _account in $user_accounts; do if [ $(sacctmgr show assoc where account=${_account} format=ParentName%100 -P | grep -v "Par Name" | grep -c courses) -eq 0 ]; then _worker_name=${HOSTNAME}-${_account} for time in 1 6 12 24 48 72; do mkdir -p ${install_path}/site_configs/${_worker_name}_${time}h && cd ${install_path}/site_configs/${_worker_name}_${time}h cat << EOF > cluster_info.json { "name": "${_worker_name}_${time}h", "worker_bin_path": "${install_path}/cryosparc_worker/bin/cryosparcw", "cache_path": "/tmp/{{ cryosparc_username }}/cryosparc_cache", "cache_reserve_mb": 10000, "cache_quota_mb": 1000000, "send_cmd_tpl": "{{ command }}", "qsub_cmd_tpl": "sbatch {{ script_path_abs }}", "qstat_cmd_tpl": "squeue -j {{ cluster_job_id }}", "qdel_cmd_tpl": "scancel {{ cluster_job_id }}", "qinfo_cmd_tpl": "sinfo" } EOF cat << EOF > cluster_script.sh #!/bin/bash -l #SBATCH --job-name cryosparc_{{ project_uid }}_{{ job_uid }} #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --cpus-per-task={{ num_cpu }} #SBATCH --gres=gpu:{{ num_gpu }} #SBATCH --mem 90GB #SBATCH --time ${time}:00:00 #SBATCH -o {{ job_dir_abs }}/slurm.out #SBATCH -e {{ job_dir_abs }}/slurm.err #SBATCH -A ${_account} -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python mkdir -p /tmp/${USER} ln -sf \${TMPDIR} /tmp/${USER}/cryosparc_cache {{ run_cmd }} EOF ${install_path}/cryosparc_master/bin/cryosparcm cluster connect done fi done for _account in $user_accounts; do if [ $(sacctmgr show assoc where account=${_account} format=ParentName%100 -P | grep -v "Par Name" | grep -c courses) -eq 0 ]; then _worker_name=${HOSTNAME}-${_account}-CPUOnly-Max20 for time in 1 6 12 24 48 72; do mkdir -p ${install_path}/site_configs/${_worker_name}_${time}h && cd ${install_path}/site_configs/${_worker_name}_${time}h cat << EOF > cluster_info.json { "name": "${_worker_name}_${time}h", "worker_bin_path": "${install_path}/cryosparc_worker/bin/cryosparcw", "cache_path": "/tmp/{{ cryosparc_username }}/cryosparc_cache", "cache_reserve_mb": 10000, "cache_quota_mb": 1000000, "send_cmd_tpl": "{{ command }}", "qsub_cmd_tpl": "sbatch {{ script_path_abs }}", "qstat_cmd_tpl": "squeue -j {{ cluster_job_id }}", "qdel_cmd_tpl": "scancel {{ cluster_job_id }}", "qinfo_cmd_tpl": "sinfo" } EOF cat << EOF > cluster_script.sh #!/bin/bash -l #SBATCH --job-name cryosparc_{{ project_uid }}_{{ job_uid }} #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --cpus-per-task={{ num_cpu }} #SBATCH --gres=gpu:1 #SBATCH --mem 90GB #SBATCH --time ${time}:00:00 #SBATCH -o {{ job_dir_abs }}/slurm.out #SBATCH -e {{ job_dir_abs }}/slurm.err #SBATCH -A ${_account} -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python mkdir -p /tmp/${USER} ln -sf \${TMPDIR} /tmp/${USER}/cryosparc_cache {{ run_cmd }} EOF ${install_path}/cryosparc_master/bin/cryosparcm cluster connect done fi done for _account in $user_accounts; do if [ $(sacctmgr show assoc where account=${_account} format=ParentName%100 -P | grep -v "Par Name" | grep -c courses) -eq 0 ]; then _worker_name=${HOSTNAME}-${_account}-SSDExclusive for time in 1 6 12 24 48 72; do mkdir -p ${install_path}/site_configs/${_worker_name}_${time}h && cd ${install_path}/site_configs/${_worker_name}_${time}h cat << EOF > cluster_info.json { "name": "${_worker_name}_${time}h", "worker_bin_path": "${install_path}/cryosparc_worker/bin/cryosparcw", "cache_path": "/tmp/{{ cryosparc_username }}/cryosparc_cache", "cache_reserve_mb": 10000, "cache_quota_mb": 1000000, "send_cmd_tpl": "{{ command }}", "qsub_cmd_tpl": "sbatch {{ script_path_abs }}", "qstat_cmd_tpl": "squeue -j {{ cluster_job_id }}", "qdel_cmd_tpl": "scancel {{ cluster_job_id }}", "qinfo_cmd_tpl": "sinfo" } EOF cat << EOF > cluster_script.sh #!/bin/bash -l #SBATCH --job-name cryosparc_{{ project_uid }}_{{ job_uid }} #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --cpus-per-task={{ num_cpu }} #SBATCH --gres=gpu:{{ num_gpu }} #SBATCH --mem 90GB #SBATCH --time ${time}:00:00 #SBATCH --exclusive #SBATCH -o {{ job_dir_abs }}/slurm.out #SBATCH -e {{ job_dir_abs }}/slurm.err #SBATCH -A ${_account} -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python mkdir -p /tmp/${USER} ln -sf \${TMPDIR} /tmp/${USER}/cryosparc_cache {{ run_cmd }} EOF ${install_path}/cryosparc_master/bin/cryosparcm cluster connect done fi done #${install_path}/cryosparc_master/bin/cryosparcm.sh stop diff --git a/create_reserved_lanes.sh b/create_reserved_lanes.sh index e406a2d..d4cf25a 100755 --- a/create_reserved_lanes.sh +++ b/create_reserved_lanes.sh @@ -1,121 +1,170 @@ #!/bin/bash set -euo pipefail # Script made from the page # https://docs.ycrc.yale.edu/clusters-at-yale/guides/cryosparc/ script_path=$(dirname "$0") # relative script_path=$(cd "${script_path}" && pwd) # absolutized and normalized install_path="$HOME"/cryosparc # set up some more paths db_path=${install_path}/database worker_path=${install_path}/cryosparc2_worker # Usage usage () { echo "Usage:" echo " -p install path : prefix for installation [${install_path}] " echo " -v : be verbose" echo " -h : print this notice" echo "" } VERBOSE=false read -p "Enter group account at SCITAS: " hpc_account read -p "Enter reservation name: " reservation_name # Parse options while getopts ":p:vh" opt; do case $opt in p) install_path="${OPTARG}" ;; v) VERBOSE=true ;; h) usage OPTIND=1 exit 0 ;; \?) echo "Invalid option: -$OPTARG" >&2 usage OPTIND=1 exit 1 ;; :) echo "Option -$OPTARG requires an argument." >&2 usage OPTIND=1 exit 1 ;; esac done # Reset OPTIND to allow the next invocation to work OPTIND=1 message() { if $VERBOSE then echo "${1}" fi } echo "[Info] Starting the master if needed" if [ $(${install_path}/cryosparc_master/bin/cryosparcm.sh status | grep -c "CryoSPARC is not running") -eq 1 ]; then ${install_path}/cryosparc_master/bin/cryosparcm.sh start fi cd ${install_path}/site_configs for _account in $hpc_account; do if [ $(sacctmgr show assoc where account=${_account} format=ParentName%100 -P | grep -v "Par Name" | grep -c courses) -eq 0 ]; then _worker_name=${HOSTNAME}-${_account}-reserved max_wall=$(sacctmgr show assoc where account=${_account} format=MaxWall -P | grep -v MaxWall | head -1) for mem in 90; do mkdir -p ${install_path}/site_configs/${_worker_name}_${mem}gb && cd ${install_path}/site_configs/${_worker_name}_${mem}gb cat << EOF > cluster_info.json { "name": "${_worker_name}_${mem}gb", "worker_bin_path": "${install_path}/cryosparc_worker/bin/cryosparcw", "cache_path": "/tmp/{{ cryosparc_username }}/cryosparc_cache", "cache_reserve_mb": 10000, "cache_quota_mb": 1000000, "send_cmd_tpl": "{{ command }}", "qsub_cmd_tpl": "sbatch {{ script_path_abs }}", "qstat_cmd_tpl": "squeue -j {{ cluster_job_id }}", "qdel_cmd_tpl": "scancel {{ cluster_job_id }}", "qinfo_cmd_tpl": "sinfo" } EOF cat << EOF > cluster_script.sh #!/bin/bash -l #SBATCH --job-name cryosparc_{{ project_uid }}_{{ job_uid }} #SBATCH --nodes=1 #SBATCH --ntasks-per-node=1 #SBATCH --cpus-per-task={{ num_cpu }} #SBATCH --gres=gpu:{{ num_gpu }} #SBATCH --mem ${mem}GB #SBATCH --time 3-00:00:00 #SBATCH -o {{ job_dir_abs }}/slurm.out #SBATCH -e {{ job_dir_abs }}/slurm.err #SBATCH -A ${_account} #SBATCH --reservation $reservation_name -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python +mkdir -p /tmp/${USER} +ln -sf \${TMPDIR} /tmp/${USER}/cryosparc_cache +{{ run_cmd }} +EOF + + ${install_path}/cryosparc_master/bin/cryosparcm cluster connect + done + fi +done + +for _account in $hpc_account; do + if [ $(sacctmgr show assoc where account=${_account} format=ParentName%100 -P | grep -v "Par Name" | grep -c courses) -eq 0 ]; then + _worker_name=${HOSTNAME}-${_account}-reserved-CPUOnly-Max20 + + + max_wall=$(sacctmgr show assoc where account=${_account} format=MaxWall -P | grep -v MaxWall | head -1) + + for mem in 90; do + mkdir -p ${install_path}/site_configs/${_worker_name}_${mem}gb && cd ${install_path}/site_configs/${_worker_name}_${mem}gb + cat << EOF > cluster_info.json +{ + "name": "${_worker_name}_${mem}gb", + "worker_bin_path": "${install_path}/cryosparc_worker/bin/cryosparcw", + "cache_path": "/tmp/{{ cryosparc_username }}/cryosparc_cache", + "cache_reserve_mb": 10000, + "cache_quota_mb": 1000000, + "send_cmd_tpl": "{{ command }}", + "qsub_cmd_tpl": "sbatch {{ script_path_abs }}", + "qstat_cmd_tpl": "squeue -j {{ cluster_job_id }}", + "qdel_cmd_tpl": "scancel {{ cluster_job_id }}", + "qinfo_cmd_tpl": "sinfo" +} +EOF + + cat << EOF > cluster_script.sh +#!/bin/bash -l +#SBATCH --job-name cryosparc_{{ project_uid }}_{{ job_uid }} +#SBATCH --nodes=1 +#SBATCH --ntasks-per-node=1 +#SBATCH --cpus-per-task={{ num_cpu }} +#SBATCH --gres=gpu:1 +#SBATCH --mem ${mem}GB +#SBATCH --time 3-00:00:00 +#SBATCH -o {{ job_dir_abs }}/slurm.out +#SBATCH -e {{ job_dir_abs }}/slurm.err +#SBATCH -A ${_account} +#SBATCH --reservation $reservation_name + +module load gcc/8.4.0 cuda/11.6.2 python mkdir -p /tmp/${USER} ln -sf \${TMPDIR} /tmp/${USER}/cryosparc_cache {{ run_cmd }} EOF ${install_path}/cryosparc_master/bin/cryosparcm cluster connect done fi done diff --git a/cryosparc_install.sh b/cryosparc_install.sh index 55b02f4..5f29ad7 100755 --- a/cryosparc_install.sh +++ b/cryosparc_install.sh @@ -1,145 +1,145 @@ #!/bin/bash set -euo pipefail # Script made from the page # https://docs.ycrc.yale.edu/clusters-at-yale/guides/cryosparc/ script_path=$(dirname "$0") # relative script_path=$(cd "${script_path}" && pwd) # absolutized and normalized license_id= port_number= install_path="$HOME"/cryosparc # set up some more paths db_path=${install_path}/database worker_path=${install_path}/cryosparc2_worker # Usage usage () { echo "Usage:" echo " -p install path : prefix for installation [${install_path}] " echo " -v : be verbose" echo " -h : print this notice" echo "" } VERBOSE=false read -p "Enter your license id: " license_id read -p "Enter the port number: " port_number # Parse options while getopts ":p:vh" opt; do case $opt in p) install_path="${OPTARG}" ;; v) VERBOSE=true ;; h) usage OPTIND=1 exit 0 ;; \?) echo "Invalid option: -$OPTARG" >&2 usage OPTIND=1 exit 1 ;; :) echo "Option -$OPTARG requires an argument." >&2 usage OPTIND=1 exit 1 ;; esac done # Reset OPTIND to allow the next invocation to work OPTIND=1 message() { if $VERBOSE then echo "${1}" fi } if [ "x${license_id}" == "x" ]; then echo "[Error] No license id was provided" >&2 usage exit 2 fi if [ "x${port_number}" == "x" ]; then echo "[Error] No part number was provided" >&2 usage exit 2 fi message "[Info] Preparing the install path and dowloading cryosparc_master and cryosparc_worker" # go get the installers mkdir -p $install_path cd $install_path # Download master curl -sL -C - -\# https://get.cryosparc.com/download/master-latest/$license_id -o cryosparc2_master.tar.gz # Download worker curl -sL -C - -\# https://get.cryosparc.com/download/worker-latest/$license_id -o cryosparc2_worker.tar.gz message "[Info] Expanding tarballs" tar -xf cryosparc2_master.tar.gz tar -xf cryosparc2_worker.tar.gz module purge -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python message "[Info] Installing master" cd ${install_path}/cryosparc_master ./install.sh --license $license_id --hostname $(hostname) --dbpath $db_path --yes set +u #source ~/.bashrc set -u message "[Info] Installing worker" cd ${install_path}/cryosparc_worker ./install.sh --license $license_id --cudapath $CUDA_HOME --yes set +u #source ~/.bashrc set -u message "[Info] Creating worker configs" cryosparc_master_path=${install_path}/cryosparc_master if [ $(grep -c "export PATH" ${cryosparc_master_path}/config.sh) -eq 0 ]; then echo "export PATH=${cryosparc_master_path}/bin:\$PATH" >> ${cryosparc_master_path}/config.sh fi if [ $(grep -c "export CRYOSPARC_HOSTNAME_CHECK" ${cryosparc_master_path}/config.sh) -eq 0 ]; then echo "export CRYOSPARC_HOSTNAME_CHECK=${HOSTNAME}" >> ${cryosparc_master_path}/config.sh fi if [ $(grep -c "export CRYOSPARC_DISABLE_IMPORT_ON_MASTER" ${cryosparc_master_path}/config.sh) -eq 0 ]; then echo "export CRYOSPARC_DISABLE_IMPORT_ON_MASTER=true" >> ${cryosparc_master_path}/config.sh fi cryosparc_worker_path=${install_path}/cryosparc_worker if [ $(grep -c "export PATH" ${cryosparc_worker_path}/config.sh) -eq 0 ]; then echo "export PATH=${cryosparc_worker_path}/bin:\$PATH" >> ${cryosparc_worker_path}/config.sh fi set +u if [ "x${port_number}" != "x" ]; then sed -i -e 's/export CRYOSPARC_BASE_PORT.*$/export CRYOSPARC_BASE_PORT='$port_number'/g' ${cryosparc_master_path}/config.sh fi set -u cp ${script_path}/cryosparcm.sh ${install_path}/cryosparc_master/bin diff --git a/cryosparcm.sh b/cryosparcm.sh index a03d44b..4dd5797 100755 --- a/cryosparcm.sh +++ b/cryosparcm.sh @@ -1,84 +1,84 @@ #!/usr/bin/env bash set -euo pipefail install_path="$HOME"/cryosparc # Usage usage () { echo "$0 -p -P command" echo "Usage:" echo " command : command to pass to cryosparcm (start|stop)" echo " -p install path : prefix for installation [${install_path}] " echo " -P base_port : base port [39000] " echo " -v : be verbose" echo " -h : print this notice" echo "" } set +u if [ "x$1" = "x" ]; then usage exit 1 fi set -u # Parse options while getopts ":P:p:h" opt; do case $opt in p) install_path="${OPTARG}" ;; P) cryosparcm_port="${OPTARG}" ;; h) usage OPTIND=1 exit 0 ;; \?) echo "Invalid option: -$OPTARG" >&2 usage OPTIND=1 exit 1 ;; :) echo "Option -$OPTARG requires an argument." >&2 usage OPTIND=1 exit 1 ;; esac done shift $((OPTIND-1)) OPTIND=1 command=$1 shift -module load gcc cuda/11.0.2 python +module load gcc/8.4.0 cuda/11.6.2 python export PATH=${install_path}/cryosparc_master/bin:$PATH cryosparcm_path="$(which cryosparcm 2> /dev/null)" base_dir=$(dirname "$(dirname "${cryosparcm_path}")") master_host=$(hostname) cp $base_dir/config.sh $base_dir/config.sh.bak sed -i -e 's/export CRYOSPARC_MASTER_HOSTNAME.*$/export CRYOSPARC_MASTER_HOSTNAME=\"'"$master_host"'\"/g' $base_dir/config.sh sed -i -e 's/export CRYOSPARC_HOSTNAME_CHECK.*$/export CRYOSPARC_HOSTNAME_CHECK=\"'"$master_host"'\"/g' $base_dir/config.sh set +u if [ "x${cryosparcm_port}" != "x" ]; then sed -i -e 's/export CRYOSPARC_BASE_PORT.*$/export CRYOSPARC_BASE_PORT='$cryosparcm_port'/g' $base_dir/config.sh fi set -u source $base_dir/config.sh ${cryosparcm_path} $command $* | sed -e 's/'$(hostname --fqdn)'/'${master_host}'.epfl.ch/' if [[ $command =~ '.*start' ]]; then ${cryosparcm_path} status fi