Browse Source

Correct failure when stopping cluster state is ignored

If the user passes in the `galera_ignore_cluster_state=true` flag, like
in the case of a node addition or a rolling replacement of nodes,
this role will fail to execute on all new nodes due to the existance of
a functioning cluster galera cluster. The first task in the pre-upgrade
steps attempts to stop the mysql service prior to running anything else.
This change makes it so the mysql service will not fail to stop on this
pre-upgrade task when the `galera_ignore_cluster_state=true` is passed
through.

Change-Id: Id9c4f11b3536e7f9d510f5ef8e6d82a82a96adac
Signed-off-by: Kevin Carter <kevin.carter@rackspace.com>
Kevin Carter 6 months ago
parent
commit
068bb72999
No account linked to committer's email address
1 changed files with 2 additions and 1 deletions
  1. 2
    1
      tasks/galera_upgrade_pre.yml

+ 2
- 1
tasks/galera_upgrade_pre.yml View File

@@ -14,9 +14,10 @@
14 14
 # limitations under the License.
15 15
 
16 16
 - name: Stop MariaDB
17
-  service:
17
+  systemd:
18 18
     name: mysql
19 19
     state: stopped
20
+  failed_when: "{{ (not galera_ignore_cluster_state | bool) | default(omit, false) }}"
20 21
   register: galera_restart_fall_back
21 22
   until: galera_restart_fall_back is success
22 23
   retries: 3

Loading…
Cancel
Save