Merge branch 'main' of gitlab.taurusgroup.one:clustervision/trinityx-…

…combined into main

README.rst

@@ -17,7 +17,7 @@ In standard configuration TrinityX provides the following services to the cluste
 * Luna, our default super-efficient node provisioner
 * OpenLDAP
 * SLURM or OpenPBS
-* Telegraf, InfluxDB, Grafana and Sensu Core
+* Prometheus and Grafana
 * Infrastructure services such as NTP, DNS, DHCP
 * and more
 
@@ -27,6 +27,7 @@ It will also set up:
 * OpenHPC applications and libraries
 * environment modules
 * rsyslog
+* High Availability
 * and more
 
 .. image:: img/triX_300.png

site/roles/trinity/luna2/tasks/main.yml

@@ -438,12 +438,34 @@
     mode: '0600'
   ignore_errors: true
 
-- name: Set me as Luna master
-  shell: "/usr/sbin/lmaster -s"
-  ignore_errors: true
+# -------------- H/A stuff ------------------
+
+- name: Render /etc/systemd/system/luna2-master.service
+  copy:
+    src: 'luna2-master.service'
+    dest: '/etc/systemd/system/luna2-master.service'
+    owner: root
+    group: root
+    mode: 0644
+  when: ha | default(False)
+
+- block:
+  - name: Set me as Luna master
+    shell: "/usr/sbin/lmaster -s"
+    ignore_errors: true
+
+  - name: Add luna2 master systemd service
+    pcs_resource:
+      name: 'luna2-master'
+      resource_type: 'systemd:luna2-master.service'
+      options: 'op monitor interval=29s --group Trinity-stack'
+      state: present
+    tags: pcs
   when:
-    - ha
-    - primary
+    - ha | default(False)
+    - primary | default(True)
+
+# ------------------------------------------
 
 - name: Setting default DNS forwarder
   shell: "luna cluster change -fs \"{{ trix_dns_forwarders|join(',') }}\""

site/roles/trinity/pacemaker/tasks/main.yaml

@@ -63,8 +63,12 @@
   template:
     src: "corosync.conf.j2"
     dest: "/etc/corosync/corosync.conf"
-  notify:
-    - restart corosync
+
+- name: Restart corosync
+  service:
+    name: corosync
+    state: restarted
+    enabled: true
 
 # TODO: better way than below
 - name: Install PCS
@@ -155,14 +159,6 @@
 
 # ---------------------- resources and config ----------------
 
-- name: Render /etc/systemd/system/luna2-master.service
-  copy:
-    src: 'luna2-master.service'
-    dest: '/etc/systemd/system/luna2-master.service'
-    owner: root
-    group: root
-    mode: 0644
-
 - block:
   - name: Set pacemaker properties
     pcs_property:
@@ -184,23 +180,6 @@
       resource_type: 'ocf:heartbeat:IPaddr2'
       options: 'ip={{ trix_ctrl_ip }} nic={{ trix_ctrl_interface }} cidr_netmask={{ trix_cluster_netprefix }} op monitor interval=29s --group Trinity'
       state: present
-
-  - name: Render temporary dummy lmaster script
-    copy:
-      src: 'lmaster'
-      dest: '/usr/sbin/lmaster'
-      owner: root
-      group: root
-      mode: 0755
-      force: no
-    ignore_errors: true
-
-  - name: Add luna2 master systemd service
-    pcs_resource:
-      name: 'luna2-master'
-      resource_type: 'systemd:luna2-master.service'
-      options: 'op monitor interval=29s --group Trinity'
-      state: present
   when: primary
 
 # --------------------- fencing ----------------------------

site/roles/trinity/shared-fs/files/lvm-fixes.sh

@@ -10,7 +10,12 @@ if [ ! "$LVM_DISKS" ] && [ ! "$VGROUPS" ]; then
 fi
 find /dev -type l -exec echo -n {}'=' \; -exec readlink -f {} \; > /tmp/dev-links.dat
 
-FILTER="["
+# Default is empty filter, but we have seen sometimes partlabel causing issues
+# where the kernel sees it at a later point, after this script ran.
+# we assume(!) that no-one is using a disk by /dev/disk/by-partlabel/primary
+# and therefor we block it from the get go.
+# FILTER = "[ "
+FILTER="[ \"r|/dev/disk/by-partlabel/primary|\","
 for DISK in $LVM_DISKS; do
 	REAL_DISK=""
 	EXCL_LINK=""

site/roles/trinity/shared-fs/tasks/main.yaml

@@ -18,6 +18,18 @@
   - debug:
       msg: "{{ drbd_fs_disks }}"
 
+  - name: Gather the disks used for DRBD
+    stat:
+      path: "{{ item.disk }}"
+    with_items: "{{ drbd_fs_disks }}"
+    register: drbd_disks_stat
+
+  - name: Verify if requested disks exist
+    fail:
+      msg: "Disk {{ item.disk }} does not seem to exist. DRBD config cannot continue, but can be re-run after enabling the disk or changing the configuration."
+    with_items: "{{ drbd_disks_stat.results | rejectattr('stat.exists') | map(attribute='item') | list }}"
+    when: drbd_disks_stat is defined
+
   - name: Configure DRBD
     ansible.builtin.include_tasks:
       file: drbd.yaml
@@ -71,6 +83,18 @@
   - debug:
       msg: "{{ direct_fs_disks }}"
 
+  - name: Gather the disks used for Direct
+    stat:
+      path: "{{ item.device }}"
+    with_items: "{{ direct_fs_disks }}"
+    register: direct_disks_stat
+
+  - name: Verify if requested devices exist
+    fail:
+      msg: "Disk {{ item.device }} does not seem to exist. Direct config cannot continue, but can be re-run after enabling the device or changing the configuration."
+    with_items: "{{ direct_disks_stat.results | rejectattr('stat.exists') | map(attribute='item') | list }}"
+    when: direct_disks_stat is defined
+
   - name: Configure Direct attached disks
     ansible.builtin.include_tasks:
       file: direct.yaml

site/roles/trinity/wrapup/tasks/main.yml

@@ -22,12 +22,17 @@
   ignore_errors: true
   when: wrapup_fix_firewalld_zones_script.stat.exists
 
-- name: set property 'standby' to 'absent' for non-primary node (unstandby)
-  pcs_property:
-    name: 'standby'
-    node: '{{ ansible_hostname }}'
-    state: 'absent'
-  ignore_errors: true
+- block:
+  - name: set property 'standby' to 'absent' for non-primary node (unstandby)
+    pcs_property:
+      name: 'standby'
+      node: '{{ ansible_hostname }}'
+      state: 'absent'
+    ignore_errors: true
+
+  - name: Clear possible pacemaker dependency messages
+    shell: "pcs resource cleanup"
+    ignore_errors: true
   tags: ha,pcs,pacemaker
   when:
     - ha