Support EESSI

.github/workflows/stackhpc.yml

@@ -99,6 +99,12 @@ jobs:
           . environments/.stackhpc/activate
           ansible-playbook -vv ansible/adhoc/hpctests.yml
 
+      - name: Run EESSI tests
+        run: |
+          . venv/bin/activate
+          . environments/.stackhpc/activate
+          ansible-playbook -vv ansible/ci/check_eessi.yml
+
       - name: Confirm Open Ondemand is up (via SOCKS proxy)
         run: |
           . venv/bin/activate

ansible/bootstrap.yml

@@ -96,6 +96,15 @@
         tasks_from: config.yml
       tags: config
 
+- name: Setup EESSI
+  hosts: eessi
+  tags: eessi
+  become: true
+  tasks:
+    - name: Install and configure EESSI
+      import_role:
+        name: eessi
+
 - hosts: update
   gather_facts: false
   become: yes

ansible/ci/check_eessi.yml

@@ -0,0 +1,26 @@
+---
+- name: Run EESSI test job
+  hosts: login # TODO: Limit this to single node when there are multiple?
+  become: true
+  become_user: testuser
+  tasks:
+    - name: Clone eessi-demo repo
+      ansible.builtin.git:
+        repo: "https://github.com/eessi/eessi-demo.git"
+        dest: ~/eessi-demo
+
+    - name: Run test job
+      ansible.builtin.shell:
+        cmd: |
+          source /cvmfs/pilot.eessi-hpc.org/latest/init/bash
+          srun ./run.sh
+        chdir: ~/eessi-demo/TensorFlow
+        executable: /bin/bash
+      register: job_output
+
+    - name: Fail if job output contains error
+      fail:
+        # Note: Job prints live progress bar to terminal, so use regex filter to remove this from stdout
+        msg: "Test job using EESSI modules failed. Job output was: {{ job_output.stdout | regex_replace('\b', '') }}"        
+      when: '"Epoch 5/5" not in job_output.stdout'
+

ansible/ci/check_sacct_hpctests.yml

@@ -2,7 +2,7 @@
   gather_facts: false
   become: true
   vars:
-    sacct_stdout_expected: |- # based on CI running hpctests as the first job - NB note no trailing newline
+    sacct_stdout_expected: |- # based on CI running hpctests as the first job
       JobID,JobName,State
       1,pingpong.sh,COMPLETED
       2,pingmatrix.sh,COMPLETED
@@ -18,10 +18,10 @@
       register: sacct
     - name: Check info for ended jobs
       assert:
-        that: sacct.stdout == sacct_stdout_expected
+        that: sacct_stdout_expected in sacct.stdout
         fail_msg: |
           Expected:
           --{{ sacct_stdout_expected }}--
           Got:
           --{{ sacct.stdout }}--
-        success_msg: sacct shows hpctests jobs as first and only jobs
+        success_msg: sacct shows hpctests jobs as first jobs in list

ansible/roles/eessi/defaults/main.yaml

@@ -0,0 +1,3 @@
+---
+# Default to 10GB
+cvmfs_quota_limit_mb: 10000

ansible/roles/eessi/tasks/main.yaml

@@ -0,0 +1,22 @@
+---
+- name: Install CVMFS repo
+  yum:
+    name: https://ecsft.cern.ch/dist/cvmfs/cvmfs-release/cvmfs-release-latest.noarch.rpm
+    # NOTE: Can't find any docs on obtaining gpg key
+    disable_gpg_check: true
+- name: Install CVMFS
+  yum:
+    name: cvmfs
+- name: Install EESSI CVMFS config
+  yum:
+    name: https://github.com/EESSI/filesystem-layer/releases/download/latest/cvmfs-config-eessi-latest.noarch.rpm
+    # NOTE: Can't find any docs on obtaining gpg key
+    disable_gpg_check: true
+- name: Add base CVMFS config
+  template:
+    src: default.local.j2
+    dest: /etc/cvmfs/default.local
+# NOTE: Not clear how to make this idempotent
+- name: Ensure CVMFS config is setup
+  command:
+    cmd: "cvmfs_config setup"

ansible/roles/eessi/templates/default.local.j2

@@ -0,0 +1,2 @@
+CVMFS_CLIENT_PROFILE="single"
+CVMFS_QUOTA_LIMIT={{ cvmfs_quota_limit_mb }}

environments/common/inventory/groups

@@ -13,6 +13,9 @@ login
 control
 compute
 
+[eessi:children]
+# Hosts on which EESSI stack should be configured
+
 [hpctests:children]
 # Login group to use for running mpi-based testing.
 login

environments/common/layouts/everything

@@ -54,3 +54,6 @@ compute
 
 [etc_hosts]
 # Hosts to manage /etc/hosts e.g. if no internal DNS. See ansible/roles/etc_hosts/README.md
+
+[eessi:children]
+openhpc