Update slurm container setup

.env

@@ -2,6 +2,15 @@
 # CCBACKEND DEVEL DOCKER SETTINGS
 ########################################################################
 
+########################################################################
+# SLURM
+########################################################################
+SLURM_VERSION=22.05.6
+ARCH=aarch64
+MUNGE_UID=981 
+SLURM_UID=982 
+WORKER_UID=1000
+
 ########################################################################
 # INFLUXDB
 ########################################################################

docker-compose.yml

@@ -52,12 +52,13 @@ services:
     command: ["--default-authentication-plugin=mysql_native_password"]
     environment:
       MARIADB_ROOT_PASSWORD: ${MARIADB_ROOT_PASSWORD}
-      MARIADB_DATABASE: ${MARIADB_DATABASE}
+      MARIADB_DATABASE: slurm_acct_db
-      MARIADB_USER: ${MARIADB_USER}
+      MARIADB_USER: slurm
-      MARIADB_PASSWORD: ${MARIADB_PASSWORD}
+      MARIADB_PASSWORD: demo
     ports:
       - "127.0.0.1:${MARIADB_PORT}:3306"
-    # volumes:
+    volumes:
+      - ${DATADIR}/mariadb:/etc/mysql/conf.d
       # - ${DATADIR}/sql-init:/docker-entrypoint-initdb.d
     cap_add:
       - SYS_NICE
@@ -81,72 +82,47 @@ services:
 
   slurm-controller:
     container_name: slurmctld
+    hostname: slurmctld
     build:
       context: ./slurm/controller
     privileged: true
     volumes:
-      - ./home:/home
+      - ${DATADIR}/slurm/home:/home
-      - ./secret:/.secret
+      - ${DATADIR}/slurm/secret:/.secret
-    restart: always
-    environment:
-      USE_SLURMDBD: 'true'
-      CLUSTER_NAME: snowflake
-      CONTROL_MACHINE: controller
-      SLURMCTLD_PORT: 6817
-      SLURMD_PORT: 6818
-      ACCOUNTING_STORAGE_HOST: database
-      ACCOUNTING_STORAGE_PORT: 6819
-      COMPUTE_NODES: worker01 worker02
-      PARTITION_NAME: docker
 
   slurm-database:
     container_name: slurmdb
+    hostname: slurmdb
     build:
       context: ./slurm/database
     depends_on:
+      - mariadb
       - slurm-controller
     privileged: true
     volumes:
-      - ./home:/home
+      - ${DATADIR}/slurm/home:/home
-      - ./secret:/.secret
+      - ${DATADIR}/slurm/secret:/.secret
-    restart: always
-    environment:
-      DBD_ADDR: database
-      DBD_HOST: database
-      DBD_PORT: 6819
-      STORAGE_HOST: database.local.dev
-      STORAGE_PORT: 3306
-      STORAGE_PASS: password
-      STORAGE_USER: slurm
 
   slurm-worker01:
     container_name: node01
+    hostname: node01
     build:
       context: ./slurm/worker
     depends_on:
       - slurm-controller
     privileged: true
     volumes:
-      - ./home:/home
+      - ${DATADIR}/slurm/home:/home
-      - ./secret:/.secret
+      - ${DATADIR}/slurm/secret:/.secret
-    restart: always
-    environment:
-      CONTROL_MACHINE: controller
-      ACCOUNTING_STORAGE_HOST: database
-      COMPUTE_NODES: worker01 worker02
 
   # slurm-worker02:
-  #   container_name: slurm-worker02
+  #   container_name: node02
+    # hostname: node02
   #   build:
   #     context: ./slurm/worker
   #   depends_on:
   #     - slurm-controller
   #   privileged: true
   #   volumes:
-  #     - ./home:/home
+      # - ${DATADIR}/slurm/home:/home
-  #     - ./secret:/.secret
+      # - ${DATADIR}/slurm/secret:/.secret
-  #   restart: always
-  #   environment:
-  #     CONTROL_MACHINE: controller
-  #     ACCOUNTING_STORAGE_HOST: database
-  #     COMPUTE_NODES: worker01 worker02

env-template.txt

@@ -1,5 +1,5 @@
-ENV SLURM_VERSION=19.05.1 \
+SLURM_VERSION=22.05.6
-  MUNGE_UID=981 \
+ARCH=aarch64
-  SLURM_UID=982 \
+MUNGE_UID=981 
-  WORKER_UID=1000
+SLURM_UID=982 
-
+WORKER_UID=1000

slurm/base/Dockerfile

@@ -18,6 +18,7 @@ RUN dnf --enablerepo=powertools install munge-devel -y
 RUN yum install rng-tools -y
 
 RUN yum install -y python3 gcc openssl openssl-devel \
+openssh-server openssh-clients dbus-devel \
 pam-devel numactl numactl-devel hwloc sudo \
 lua readline-devel ncurses-devel man2html \
 libibmad libibumad rpm-build  perl-ExtUtils-MakeMaker.noarch rpm-build make wget
@@ -42,4 +43,4 @@ VOLUME ["/home", "/.secret"]
 # 6817:         SlurmCtlD
 # 6818:         SlurmD
 # 6819:         SlurmDBD
-EXPOSE 22 3306 6817 6818 6819
+EXPOSE 22 6817 6818 6819

slurm/base/Makefile

@@ -1,4 +1,5 @@
-IMAGE = scidas/slurm.base
+include ../../.env
+IMAGE = clustercockpit/slurm.base
 
 .PHONY: build clean
 

slurm/controller/Dockerfile

@@ -1,4 +1,4 @@
-FROM clustercockpit/slurm.base:latest
+FROM clustercockpit/slurm.base:22.05.6
 MAINTAINER Jan Eitzinger <jan.eitzinger@fau.de>
 
 # clean up

slurm/controller/docker-entrypoint.sh

@@ -7,6 +7,7 @@ _sshd_host() {
     mkdir /var/run/sshd
     ssh-keygen -t rsa -f /etc/ssh/ssh_host_rsa_key -N ''
   fi
+  echo "Starting sshd"
   /usr/sbin/sshd
 }
 
@@ -40,6 +41,7 @@ EOF2
 
 # start munge and generate key
 _munge_start() {
+    echo  "Starting munge"
     chown -R munge: /etc/munge /var/lib/munge /var/log/munge /var/run/munge
     chmod 0700 /etc/munge
     chmod 0711 /var/lib/munge
@@ -69,8 +71,12 @@ _copy_secrets() {
 # run slurmctld
 _slurmctld() {
     cd /root/rpmbuild/RPMS/aarch64
-    yum -y --nogpgcheck localinstall slurm-22.05.6-1.el8.aarch64.rpm slurm-perlapi-22.05.6-1.el8.aarch64.rpm  slurm-slurmctld-22.05.6-1.el8.aarch64.rpm
+    yum -y --nogpgcheck localinstall slurm-22.05.6-1.el8.aarch64.rpm \
-    echo -n "checking for slurmdbd.conf"
+        slurm-perlapi-22.05.6-1.el8.aarch64.rpm \
+        slurm-slurmd-22.05.6-1.el8.aarch64.rpm \
+        slurm-torque-22.05.6-1.el8.aarch64.rpm \
+        slurm-slurmctld-22.05.6-1.el8.aarch64.rpm
+    echo "checking for slurmdbd.conf"
     while [ ! -f /.secret/slurmdbd.conf ]; do
         echo -n "."
         sleep 1
@@ -86,11 +92,14 @@ _slurmctld() {
     else
         echo "### use provided slurm.conf ###"
         cp /home/config/slurm.conf /etc/slurm/slurm.conf
+        chown slurm: /etc/slurm/slurm.conf
+        chmod 600 /etc/slurm/slurm.conf
     fi
     sacctmgr -i add cluster "snowflake"
     sleep 2s
-    /usr/sbin/slurmctld
+    echo  "Starting slurmctld"
     cp -f /etc/slurm/slurm.conf /.secret/
+    /usr/sbin/slurmctld
 }
 
 ### main ###

slurm/database/Dockerfile

@@ -1,4 +1,4 @@
-FROM clustercockpit/slurm.base:latest
+FROM clustercockpit/slurm.base:22.05.6
 MAINTAINER Jan Eitzinger <jan.eitzinger@fau.de>
 
 # clean up

slurm/database/docker-entrypoint.sh

@@ -12,32 +12,6 @@ _sshd_host() {
   /usr/sbin/sshd
 }
 
-# slurm database user settings
-_slurm_acct_db() {
-  {
-    echo "create database slurm_acct_db;"
-    echo "create user 'slurm'@slurmdb'';"
-    echo "set password for 'slurm'@'slurmdb' = password('demo');"
-    echo "grant usage on *.* to 'slurm'@'slurmdb';"
-    echo "grant all privileges on slurm_acct_db.* to 'slurm'@'slurmdb';"
-    echo "flush privileges;"
-  } >> $SLURM_ACCT_DB_SQL
-}
-
-# start database
-_mariadb_start() {
-  # mariadb somehow expects `resolveip` to be found under this path; see https://github.com/SciDAS/slurm-in-docker/issues/26
-  ln -s /usr/bin/resolveip /usr/libexec/resolveip
-  mysql_install_db
-  chown -R mysql: /var/lib/mysql/ /var/log/mariadb/ /var/run/mariadb
-  cd /var/lib/mysql
-  mysqld_safe --user=mysql &
-  cd /
-  _slurm_acct_db
-  sleep 5s
-  mysql -uroot < $SLURM_ACCT_DB_SQL
-}
-
 # start munge using existing key
 _munge_start_using_key() {
   if [ ! -f /.secret/munge.key ]; then
@@ -74,24 +48,28 @@ _wait_for_worker() {
 
 # run slurmdbd
 _slurmdbd() {
-  mkdir -p /var/spool/slurm/d \
+   cd /root/rpmbuild/RPMS/aarch64
-    /var/log/slurm
+   yum -y --nogpgcheck localinstall slurm-22.05.6-1.el8.aarch64.rpm \
-  chown slurm: /var/spool/slurm/d \
+       slurm-perlapi-22.05.6-1.el8.aarch64.rpm \
-    /var/log/slurm
+       slurm-slurmdbd-22.05.6-1.el8.aarch64.rpm
+  mkdir -p /var/spool/slurm/d /var/log/slurm /etc/slurm
+  chown slurm: /var/spool/slurm/d /var/log/slurm
   if [[ ! -f /home/config/slurmdbd.conf ]]; then
     echo "### Missing slurmdbd.conf ###"
     exit
   else
     echo "### use provided slurmdbd.conf ###"
     cp /home/config/slurmdbd.conf /etc/slurm/slurmdbd.conf
+    chown slurm: /etc/slurm/slurmdbd.conf
+    chmod 600 /etc/slurm/slurmdbd.conf
   fi
-  /usr/sbin/slurmdbd
+  echo "Starting slurmdbd"
   cp /etc/slurm/slurmdbd.conf /.secret/slurmdbd.conf
+  /usr/sbin/slurmdbd
 }
 
 ### main ###
 _sshd_host
-_mariadb_start
 _munge_start_using_key
 _wait_for_worker
 _slurmdbd

slurm/rest/Dockerfile

@@ -0,0 +1,10 @@
+FROM clustercockpit/slurm.base:22.05.6
+MAINTAINER Jan Eitzinger <jan.eitzinger@fau.de>
+
+# clean up
+RUN rm -f /root/rpmbuild/RPMS/slurm-*.rpm \
+  && yum clean all \
+  && rm -rf /var/cache/yum
+
+COPY docker-entrypoint.sh /docker-entrypoint.sh
+ENTRYPOINT ["/docker-entrypoint.sh"]

slurm/rest/docker-entrypoint.sh

@@ -0,0 +1,108 @@
+#!/usr/bin/env bash
+set -e
+
+# start sshd server
+_sshd_host() {
+  if [ ! -d /var/run/sshd ]; then
+    mkdir /var/run/sshd
+    ssh-keygen -t rsa -f /etc/ssh/ssh_host_rsa_key -N ''
+  fi
+  /usr/sbin/sshd
+}
+
+# setup worker ssh to be passwordless
+_ssh_worker() {
+    if [[ ! -d /home/worker  ]]; then
+        mkdir -p /home/worker
+        chown -R worker:worker /home/worker
+    fi
+    cat > /home/worker/setup-worker-ssh.sh <<EOF2
+mkdir -p ~/.ssh
+chmod 0700 ~/.ssh
+ssh-keygen -b 2048 -t rsa -f ~/.ssh/id_rsa -q -N "" -C "$(whoami)@$(hostname)-$(date -I)"
+cat ~/.ssh/id_rsa.pub > ~/.ssh/authorized_keys
+chmod 0640 ~/.ssh/authorized_keys
+cat >> ~/.ssh/config <<EOF
+Host *
+StrictHostKeyChecking no
+UserKnownHostsFile /dev/null
+LogLevel QUIET
+EOF
+chmod 0644 ~/.ssh/config
+cd ~/
+tar -czvf ~/worker-secret.tar.gz .ssh
+cd -
+EOF2
+    chmod +x /home/worker/setup-worker-ssh.sh
+    chown worker: /home/worker/setup-worker-ssh.sh
+    sudo -u worker /home/worker/setup-worker-ssh.sh
+}
+
+# start munge and generate key
+_munge_start() {
+    chown -R munge: /etc/munge /var/lib/munge /var/log/munge /var/run/munge
+    chmod 0700 /etc/munge
+    chmod 0711 /var/lib/munge
+    chmod 0700 /var/log/munge
+    chmod 0755 /var/run/munge
+    /sbin/create-munge-key -f
+    rngd -r /dev/urandom
+    /usr/sbin/create-munge-key -r -f
+    sh -c  "dd if=/dev/urandom bs=1 count=1024 > /etc/munge/munge.key"
+    chown munge: /etc/munge/munge.key
+    chmod 400 /etc/munge/munge.key
+    sudo -u munge /sbin/munged
+    munge -n
+    munge -n | unmunge
+    remunge
+}
+
+# copy secrets to /.secret directory for other nodes
+_copy_secrets() {
+  cp /home/worker/worker-secret.tar.gz /.secret/worker-secret.tar.gz
+  cp thome/worker/setup-worker-ssh.sh /.secret/setup-worker-ssh.sh
+  cp /etc/munge/munge.key /.secret/munge.key
+  rm -f /home/worker/worker-secret.tar.gz
+  rm -f /home/worker/setup-worker-ssh.sh
+}
+
+# run slurmctld
+_slurmctld() {
+    cd /root/rpmbuild/RPMS/aarch64
+    yum -y --nogpgcheck localinstall slurm-22.05.6-1.el8.aarch64.rpm \
+        slurm-perlapi-22.05.6-1.el8.aarch64.rpm \
+        slurm-slurmd-22.05.6-1.el8.aarch64.rpm \
+        slurm-torque-22.05.6-1.el8.aarch64.rpm \
+        slurm-slurmctld-22.05.6-1.el8.aarch64.rpm \
+        slurm-slurmrestd-22.05.6-1.el8.aarch64.rpm
+    echo -n "checking for slurmdbd.conf"
+    while [ ! -f /.secret/slurmdbd.conf ]; do
+        echo -n "."
+        sleep 1
+    done
+    echo ""
+    mkdir -p /var/spool/slurm/ctld /var/spool/slurm/d  /var/log/slurm /etc/slurm
+    chown -R slurm: /var/spool/slurm/ctld /var/spool/slurm/d  /var/log/slurm
+    touch /var/log/slurmctld.log
+    chown slurm: /var/log/slurmctld.log
+    if [[ ! -f /home/config/slurm.conf ]]; then
+        echo "### Missing slurm.conf ###"
+        exit
+    else
+        echo "### use provided slurm.conf ###"
+        cp /home/config/slurm.conf /etc/slurm/slurm.conf
+    fi
+    sacctmgr -i add cluster "snowflake"
+    sleep 2s
+    /usr/sbin/slurmctld
+    cp -f /etc/slurm/slurm.conf /.secret/
+}
+
+### main ###
+_sshd_host
+_ssh_worker
+_munge_start
+_copy_secrets
+_slurmctld
+
+tail -f /dev/null

slurm/worker/Dockerfile

@@ -1,4 +1,4 @@
-FROM clustercockpit/slurm.base:latest
+FROM clustercockpit/slurm.base:22.05.6
 MAINTAINER Jan Eitzinger <jan.eitzinger@fau.de>
 
 # clean up

slurm/worker/docker-entrypoint.sh

@@ -33,7 +33,7 @@ _munge_start_using_key() {
 # wait for worker user in shared /home volume
 _wait_for_worker() {
   if [ ! -f /home/worker/.ssh/id_rsa.pub ]; then
-    echo -n "cheking for id_rsa.pub"
+    echo -n "checking for id_rsa.pub"
     while [ ! -f /home/worker/.ssh/id_rsa.pub ]; do
       echo -n "."
       sleep 1
@@ -42,28 +42,46 @@ _wait_for_worker() {
   fi
 }
 
+_start_dbus() {
+    dbus-uuidgen > /var/lib/dbus/machine-id
+    mkdir -p /var/run/dbus
+    dbus-daemon --config-file=/usr/share/dbus-1/system.conf --print-address
+}
+
 # run slurmd
 _slurmd() {
-  if [ ! -f /.secret/slurm.conf ]; then
+    cd /root/rpmbuild/RPMS/aarch64
-    echo -n "checking for slurm.conf"
+    yum -y --nogpgcheck localinstall slurm-22.05.6-1.el8.aarch64.rpm \
-    while [ ! -f /.secret/slurm.conf ]; do
+        slurm-perlapi-22.05.6-1.el8.aarch64.rpm \
-      echo -n "."
+        slurm-slurmd-22.05.6-1.el8.aarch64.rpm \
-      sleep 1
+        slurm-torque-22.05.6-1.el8.aarch64.rpm
-    done
+    if [ ! -f /.secret/slurm.conf ]; then
-    echo ""
+        echo -n "checking for slurm.conf"
-  fi
+        while [ ! -f /.secret/slurm.conf ]; do
-  mkdir -p /var/spool/slurm/d
+          echo -n "."
-  chown slurm: /var/spool/slurm/d
+          sleep 1
-  cp /.secret/slurm.conf /etc/slurm/slurm.conf
+        done
-  touch /var/log/slurmd.log
+        echo ""
-  chown slurm: /var/log/slurmd.log
+    fi
-  /usr/sbin/slurmd
+    mkdir -p /var/spool/slurm/d /etc/slurm
+    chown slurm: /var/spool/slurm/d
+    cp /home/config/cgroup.conf /etc/slurm/cgroup.conf
+    chown slurm: /etc/slurm/cgroup.conf
+    chmod 600 /etc/slurm/cgroup.conf
+    cp /home/config/slurm.conf /etc/slurm/slurm.conf
+    chown slurm: /etc/slurm/slurm.conf
+    chmod 600 /etc/slurm/slurm.conf
+    touch /var/log/slurmd.log
+    chown slurm: /var/log/slurmd.log
+    echo -n "Starting slurmd"
+    /usr/sbin/slurmd
 }
 
 ### main ###
 _sshd_host
 _munge_start_using_key
 _wait_for_worker
+_start_dbus
 _slurmd
 
 tail -f /dev/null