openstack/kolla-ansible
Code Issues Proposed changes

Make it possible to not haproxy memcached

Browse Source 
Clients usually have sophisticated logic for handling
key redistribution on node failure; so going under the
covers and messing with that is bad (because then the
clients and their sophisticated logic is unaware of
what is going on).

Change-Id: Ica12240440d28f930b917d5d6202f4f9e6675b2a
This commit is contained in:
Joshua Harlow 2018-03-27 10:02:22 -07:00
parent e66cb5d46b
commit e82fb95603
2 changed files with 7 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
ansible/group_vars/all.yml
View File

@ -375,6 +375,12 @@ enable_nova: "yes"
enable_rabbitmq: "{{ 'yes' if om_rpc_transport == 'rabbit' or om_notify_transport == 'rabbit' else 'no' }}"
enable_outward_rabbitmq: "{{ enable_murano | bool }}"
# Most memcache clients handle load-balancing via client side
# hashing (consistent or not) logic, so going under the covers and messing
# with things that the clients are not aware of is general wrong (but this
# keeps the default as is...)
enable_haproxy_memcached: "yes"
# Additional optional OpenStack features and services are specified here
enable_aodh: "no"
enable_barbican: "no"

2
ansible/roles/haproxy/templates/haproxy.cfg.j2
View File

@ -921,7 +921,7 @@ listen mariadb
{% endfor %}
{% endif %}
{% if enable_memcached | bool %}
{% if enable_memcached | bool and enable_haproxy_memcached | bool %}
listen memcached
mode tcp
timeout client 3600s