Deploy CoD cluster

assigned to @louistw

marked this issue as related to #261 (closed)

Adding previous instances of this issue from previous sprints.

We'll keep this issue running and move them across sprints to maintain a history

Noticed shibboleth configuration issue when deploying with samltest idp.

Further investigation shows that the tarball for smaltest idp does not include the key and cert. Therefore each deployment uses a different key and cert. Since we don't update Metadata every time, it could not find the key to decrypt the message.

Solution:

Update the tarball to include key and cert, and upload the latest metadata, which has the included key and cert, to samltest.id.

Tarball has updated to include key and cert. Tested with samltest.id and worked fine.

added Sprint 23-04 label and removed Sprint 23-03 label

Deployed today, found issue in enable_lmod role, #258 (comment 74063), other than that, SSO with samltest.id works fine.

added Sprint 23-05 label and removed Sprint 23-04 label

assigned to @krish94 and unassigned @louistw

changed the description

added Sprint 23-06 label and removed Sprint 23-05 label

assigned to @ravi89 and unassigned @krish94

Sprint 23-06

Ran into an error while trying to build CoD:

TASK [enable_lmod : Make sure lmod config is installed with cm version] *********************
fatal: [ohpc]: FAILED! => {"changed": false, "changes": {"installed": ["cm-modules-init-client"]}, "msg": "\n\nTransaction check error:\n  file /etc/cm-release from install of cm-slave-9.0-44_cm9.0.noarch conflicts with file from package cm-master-9.0-49_cm9.0.noarch\n  file /etc/profile.d/modules.sh from install of cm-modules-init-client-9.0-69_cm9.0.noarch conflicts with file from package cm-modules-init-9.0-64_cm9.0.noarch\n\nError Summary\n-------------\n\n", "rc": 1, "results": ["Loaded plugins: fastestmirror, priorities\nLoading mirror speeds from cached hostfile\n * base: mirror.team-cymru.com\n * cm-ml-rhel7-9.0-updates: updates-us-east.brightcomputing.com\n * cm-rhel7-9.0-updates: updates-us-east.brightcomputing.com\n * epel: ftp-nyc.osuosl.org\n * extras: mirrors.greenmountainaccess.net\n * updates: distro.ibiblio.org\n115 packages excluded due to repository priority protections\nResolving Dependencies\n--> Running transaction check\n---> Package cm-modules-init-client.noarch 0:9.0-69_cm9.0 will be installed\n--> Processing Dependency: cm-slave for package: cm-modules-init-client-9.0-69_cm9.0.noarch\n--> Running transaction check\n---> Package cm-slave.noarch 0:9.0-44_cm9.0 will be installed\n--> Finished Dependency Resolution\n\nDependencies Resolved\n\n================================================================================\n Package                  Arch     Version         Repository              Size\n================================================================================\nInstalling:\n cm-modules-init-client   noarch   9.0-69_cm9.0    cm-rhel7-9.0-updates   6.4 k\nInstalling for dependencies:\n cm-slave                 noarch   9.0-44_cm9.0    cm-rhel7-9.0-updates   2.4 k\n\nTransaction Summary\n================================================================================\nInstall  1 Package (+1 Dependent package)\n\nTotal size: 8.8 k\nInstalled size: 8.8 k\nDownloading packages:\nRunning transaction check\nRunning transaction test\n"]}

PLAY RECAP **********************************************************************************
ohpc                       : ok=2    changed=0    unreachable=0    failed=1    skipped=0    rescued=0    ignored=0

Looks like a conflict between versions of cm

Here's the ansible role where the error occurs:

- name: Tasks only needed in cm version
  block:
    - name: Make sure lmod config is installed with cm version
      ansible.builtin.yum:
        name: cm-modules-init-client
        state: present

    - name: Enable Lmod
      replace:
        path: "{{ enable_lmod_prefix }}/etc/sysconfig/modules/lmod/{{ item.path }}"
        regexp: "{{ item.regexp }}"
        replace: "{{ item.replace }}"
      loop:
        - { path: 'cm-lmod-init.sh', regexp: 'LMOD=.*$', replace: 'LMOD=1'}
        - { path: 'cm-lmod-init.csh', regexp: 'LMOD.*$', replace: 'LMOD "1"'}

    - name: Update Lmod Spider Cache setting
      ansible.builtin.replace:
        path: "{{ enable_lmod_prefix }}{{ lmod_loc }}/init/lmodrc.lua"
        regexp: '(\["{{ item.key }}"\] = ).*$'
        replace: \1 "{{ item.value }}",
      loop:
        - {"key": "dir", "value": "{{ lmod_cache_loc }}/sysCacheDir"}
        - {"key": "timestamp", "value": "{{ lmod_cache_loc }}/sysCacheTS.txt"}
  when: '"cm" in ansible_facts.packages.Lmod[0].release'

Issue was fixed in https://github.com/jprorama/CRI_XCBC/pull/423

Tested and merged

added Sprint 23-07 label and removed Sprint 23-06 label

assigned to @atlurie and unassigned @ravi89

added Sprint 23-08 label and removed Sprint 23-07 label

added Sprint 23-09 label and removed Sprint 23-08 label

added Sprint 23-10 label and removed Sprint 23-09 label

Deploy CoD cluster

Designs

Child items 0

Activity