Fix Horizon crash on Memcached node failure

Horizon uses Django's PyMemcacheCache backend
to store session data.

When one of the Memcached nodes becomes unavailable,
Django raises an exception during session access,
causing Horizon to return HTTP 500 errors.

This patch adds 'ignore_exc': True to the cache OPTIONS
to make Django treat such errors as cache misses
instead of crashing.

This improves Horizon stability in high-availability
setups with multiple Memcached nodes.

Closes-Bug: #2106557
Change-Id: If8873eaec35c4d2e5b13cdb8bcefb96230c59b51
Signed-off-by: Michal Arbet <[email protected]>

Author: keuko

ansible/roles/horizon/templates/_9998-kolla-settings.py.j2

@@ -19,6 +19,7 @@ DATABASES = {
 {% elif groups['memcached'] | length > 0 and not horizon_backend_database | bool %}
 SESSION_ENGINE = 'django.contrib.sessions.backends.cache'
 CACHES['default']['LOCATION'] = [{% for host in groups['memcached'] %}'{{ 'api' | kolla_address(host) | put_address_in_context('url') }}:{{ memcached_port }}'{% if not loop.last %},{% endif %}{% endfor %}]
+CACHES['default']['OPTIONS'] = {'ignore_exc': True}
 {% endif %}
 
 {% if kolla_enable_tls_external | bool or kolla_enable_tls_internal | bool %}

releasenotes/notes/bug-2106557-6adff0f76b17500e.yaml

@@ -0,0 +1,7 @@
+---
+fixes:
+  - |
+    Fixes an issue where Horizon returned HTTP 500 errors when one of the
+    Memcached nodes was unavailable by setting ``ignore_exc`` to ``True`` in
+    the cache backend.
+    `LP#2106557 <https://bugs.launchpad.net/kolla-ansible/+bug/2106557>`__