slurm startup

roles/slurm/files/slurmctld.Dockerfile

@@ -16,4 +16,5 @@ RUN chown root:root /usr/local/sbin/entrypoint.sh && chmod 755 /usr/local/sbin/e
 
 ENTRYPOINT [ "/usr/local/sbin/entrypoint.sh" ]
 
-# CMD bash -c 'cat <({ condor_master -f & tail --retry --pid $! -f /var/log/condor/MasterLog & })'
+CMD bash -c 'cat <({ su -s /bin/sh -c "munged -F" munge & slurmctld -D & })'
+# ... & tail --retry --pid $! -f /var/log/condor/MasterLog & })'

roles/slurm/files/slurmd.Dockerfile

@@ -14,3 +14,4 @@ RUN chown root:root /usr/local/sbin/entrypoint.sh && chmod 755 /usr/local/sbin/e
 ENTRYPOINT [ "/usr/local/sbin/entrypoint.sh" ]
 
 # CMD bash -c 'cat <({ condor_master -f & tail --retry --pid $! -f /var/log/condor/MasterLog & })'
+CMD bash -c 'cat <({ su -s /bin/sh -c "munged -F" munge & slurmd -D & })'

roles/slurm/tasks/main.yml

@@ -34,3 +34,56 @@
     mode: preserve
     src: /etc/munge/munge.key
     dest: /container/volumes/munge/munge.key
+
+- file:
+    path: /container/volumes/slurm/
+    state: directory
+
+- name: copy slurm config
+  copy:
+    force: true
+    src: "{{item}}"
+    dest: "/container/volumes/slurm/{{item}}"
+  loop:
+    - slurm.conf
+    - cgroup.conf
+  tags: [ slurm-config ]
+
+- name: "create docker network to make service discovery work"
+  docker_network:
+    name: slurm
+    state: present
+
+# TODO: reserve some address using docker_network_info and assign as aux
+# address to enable slurmctld to get a static address in order to be
+# reachable from slurm running on docker host to enable submitting jobs.
+
+- name: run slurm docker containers
+  docker_container:
+    name: "slurm-{{item.machine}}"
+    hostname: "slurm-{{item.machine}}"
+    domainname: "slurm.local"
+    volumes: "{{default_mounts + ( item.extra_mounts | default([]) ) }}"
+    networks:
+    - name: "slurm"
+    image: "{{item.image}}"
+    state: started
+    detach: True
+    cleanup: True
+    networks_cli_compatible: True
+  vars:
+    default_mounts:
+    - /container/volumes/slurm/:/etc/slurm/:rw
+    - /container/volumes/munge/munge.key:/etc/munge/munge.key:rw
+  loop:
+  - machine: ctl
+    image: slurm-slurmctld
+  - machine: exec1
+    image: slurm-slurmd
+  - machine: exec2
+    image: slurm-slurmd
+  - machine: submit1
+    image: slurm-slurmd
+    extra_mounts:
+    - "/home/thoto/job3/:/mnt/:rw"
+  tags: [ slurm-config ]