shrink-osd: Stop ceph-disk container based on ID

Since bedc0ab we now manage ceph-osd systemd unit scripts based on ID
instead of device name but it was not present in the shrink-osd
playbook (ceph-disk version).
To keep backward compatibility on deployment that didn't do yet the
transition on OSD id then we should stop unit scripts for both device
and ID.
This commit adds the ulimit nofile container option to get better
performance on ceph-disk commands.
It also fixes an issue when the OSD id matches multiple OSD ids with
the same first digit.

$ ceph-disk list | grep osd.1
 /dev/sdb1 ceph data, prepared, cluster ceph, osd.1, block /dev/sdb2
 /dev/sdg1 ceph data, prepared, cluster ceph, osd.12, block /dev/sdg2

Finally removing the shrinked OSD directory.

Signed-off-by: Dimitri Savineau <dsavinea@redhat.com>

diff --git a/infrastructure-playbooks/shrink-osd-ceph-disk.yml b/infrastructure-playbooks/shrink-osd-ceph-disk.yml
index c889f5f99e2dc9e788855ab3deeb022dc10778ac..8059e2e8e320256b690d29d7d960fa4d6103e6e3 100644 (file)
--- a/infrastructure-playbooks/shrink-osd-ceph-disk.yml
+++ b/infrastructure-playbooks/shrink-osd-ceph-disk.yml
@@ -106,9 +106,9 @@
     # NOTE(leseb): using '>' is the only way I could have the command working
     - name: find osd device based on the id
       shell: >
-        {{ 'docker run --privileged=true -v /dev:/dev --entrypoint' if containerized_deployment else '' }} /usr/sbin/ceph-disk
+        {{ 'docker run --privileged=true --ulimit nofile=1024:4096 -v /dev:/dev --entrypoint' if containerized_deployment else '' }} /usr/sbin/ceph-disk
         {{ ceph_docker_registry + '/' + ceph_docker_image + ':' + ceph_docker_image_tag if containerized_deployment else '' }}
-        list | awk  -v pattern=osd.{{ item.0 }} '$0 ~ pattern {print $1}'
+        list | awk  -v pattern=osd.{{ item.0 }}, '$0 ~ pattern {print $1}'
       with_together:
         - "{{ osd_to_kill.split(',') }}"
         - "{{ osd_hosts }}"
@@ -117,9 +117,9 @@
 
     - name: find osd dedicated devices - container
       shell: >
-        docker run --privileged=true -v /dev:/dev --entrypoint /usr/sbin/ceph-disk
+        docker run --privileged=true --ulimit nofile=1024:4096 -v /dev:/dev --entrypoint /usr/sbin/ceph-disk
         {{ ceph_docker_registry}}/{{ ceph_docker_image }}:{{ ceph_docker_image_tag }}
-        list | grep osd.{{ item.0 }} | grep -Eo '/dev/([hsv]d[a-z]{1,2})[0-9]{1,2}|/dev/nvme[0-9]n[0-9]p[0-9]'
+        list | grep osd.{{ item.0 }}, | grep -Eo '/dev/([hsv]d[a-z]{1,2})[0-9]{1,2}|/dev/nvme[0-9]n[0-9]p[0-9]'
       with_together:
         - "{{ osd_to_kill.split(',') }}"
         - "{{ osd_hosts }}"
@@ -129,7 +129,7 @@
         - containerized_deployment
 
     - name: find osd dedicated devices - non container
-      shell: ceph-disk list | grep osd.{{ item.0 }} | grep -Eo '/dev/([hsv]d[a-z]{1,2})[0-9]{1,2}|/dev/nvme[0-9]n[0-9]p[0-9]'
+      shell: ceph-disk list | grep osd.{{ item.0 }}, | grep -Eo '/dev/([hsv]d[a-z]{1,2})[0-9]{1,2}|/dev/nvme[0-9]n[0-9]p[0-9]'
       with_together:
         - "{{ osd_to_kill.split(',') }}"
         - "{{ osd_hosts }}"
@@ -140,7 +140,7 @@
 
     # if nvme then osd_to_kill_disks is nvme0n1, we need nvme0
     # if ssd or hdd then osd_to_kill_disks is sda1, we need sda
-    - name: stop osd services (container)
+    - name: stop osd services with device (container)
       service:
         name: "ceph-osd@{{ item.0.stdout[:-2] | regex_replace('/dev/', '') if 'nvme' in item.0.stdout else item.0.stdout[:-1] | regex_replace('/dev/', '') }}"
         state: stopped
@@ -152,6 +152,17 @@
       when:
         - containerized_deployment
 
+    - name: stop osd services with ID (container)
+      service:
+        name: "ceph-osd@{{ item.0 }}"
+        state: stopped
+        enabled: no
+      with_together:
+        - "{{ osd_to_kill.split(',') }}"
+        - "{{ osd_hosts }}"
+      delegate_to: "{{ item.1 }}"
+      when: containerized_deployment | bool
+
     - name: resolve parent device
       command: lsblk --nodeps -no pkname "{{ item.0.stdout }}"
       register: resolved_parent_device
@@ -278,6 +289,15 @@
         - not containerized_deployment
         - item.1 | length > 0
 
+    - name: remove ceph osd directory
+      file:
+        name: '/var/lib/ceph/osd/{{ cluster }}-{{ item.0 }}'
+        state: absent
+      with_together:
+        - "{{ osd_to_kill.split(',') }}"
+        - "{{ osd_hosts }}"
+      delegate_to: "{{ item.1 }}"
+
     - name: show ceph health
       command: "{{ docker_exec_cmd }} ceph --cluster {{ cluster }} -s"
       delegate_to: "{{ groups[mon_group_name][0] }}"