Merge pull request #664 from stackhpc/upstream/master-2025-01-23

Synchronise master with upstream

Author: priteau

ansible/group_vars/all.yml

@@ -529,27 +529,6 @@ mariadb_clustercheck_port: "4569"
 mariadb_enable_tls_backend: "{{ database_enable_tls_backend }}"
 
 mariadb_monitor_user: "{{ 'monitor' if enable_proxysql | bool else 'haproxy' }}"
-mariadb_monitor_connect_interval: "2000"
-mariadb_monitor_galera_healthcheck_interval: "4000"
-mariadb_monitor_galera_healthcheck_timeout: "1000"
-mariadb_monitor_galera_healthcheck_max_timeout_count: "2"
-mariadb_monitor_ping_interval: "3000"
-mariadb_monitor_ping_timeout: "2000"
-mariadb_monitor_ping_max_failures: "2"
-#
-# Defaults preserved for multinode setup
-# Tweaked for single-node
-#
-# ProxySQL shuns servers on MySQL errors, which can cause failures
-# during upgrades or restarts. In single-node setups, ProxySQL can't reroute
-# traffic, leading to "Max connect timeout" errors. To avoid this in CI and
-# signle-node environments, delay error responses to clients by 10 seconds,
-# giving the backend time to recover without immediate failures.
-#
-# See ProxySQL docs for more: https://proxysql.com/documentation/global-variables/mysql-variables/#mysql-shun_on_failures
-mariadb_shun_on_failures: "{{ '10' if mariadb_shards_info.shards.values() | map(attribute='hosts') | map('length') | select('<=', 1) | list | length > 0 else '5' }}"
-mariadb_connect_retries_delay: "{{ '1000' if mariadb_shards_info.shards.values() | map(attribute='hosts') | map('length') | select('<=', 1) | list | length > 0 else '1' }}"
-mariadb_connect_retries_on_failure: "{{ '20' if mariadb_shards_info.shards.values() | map(attribute='hosts') | map('length') | select('<=', 1) | list | length > 0 else '10' }}"
 
 mariadb_datadir_volume: "mariadb"
 

ansible/roles/loadbalancer/defaults/main.yml

@@ -205,4 +205,31 @@ loadbalancer_copy_certs: "{{ kolla_copy_ca_into_containers | bool or kolla_enabl
 ################
 # ProxySQL
 ################
+mariadb_connect_timeout_client: ""
+mariadb_connect_timeout_server: ""
+mariadb_connect_timeout_server_max: ""
+# ProxySQL default is two minutes, which is most probably too much.
+mariadb_monitor_connect_interval: "2000"
+mariadb_monitor_connect_timeout: ""
+mariadb_monitor_galera_healthcheck_interval: ""
+mariadb_monitor_galera_healthcheck_timeout: ""
+mariadb_monitor_galera_healthcheck_max_timeout_count: ""
+mariadb_monitor_ping_interval: ""
+mariadb_monitor_ping_timeout: ""
+mariadb_monitor_ping_max_failures: ""
 mariadb_monitor_read_only_interval: ""
+#
+# Defaults preserved for multinode setup
+# Tweaked for single-node
+#
+# ProxySQL shuns servers on MySQL errors, which can cause failures
+# during upgrades or restarts. In single-node setups, ProxySQL can't reroute
+# traffic, leading to "Max connect timeout" errors. To avoid this in CI and
+# signle-node environments, delay error responses to clients by 10 seconds,
+# giving the backend time to recover without immediate failures.
+#
+# See ProxySQL docs for more: https://proxysql.com/documentation/global-variables/mysql-variables/#mysql-shun_on_failures
+mariadb_singlenode: "{{ mariadb_shards_info.shards.values() | map(attribute='hosts') | map('length') | select('<=', 1) | list | length > 0 }}"
+mariadb_shun_on_failures: "{{ '10' if mariadb_singlenode else '' }}"
+mariadb_connect_retries_delay: "{{ '1000' if mariadb_singlenode else '' }}"
+mariadb_connect_retries_on_failure: "{{ '20' if mariadb_singlenode else '' }}"

ansible/roles/loadbalancer/templates/proxysql/proxysql.yaml.j2

@@ -20,25 +20,53 @@ mysql_variables:
    threads: {{ proxysql_workers }}
    max_connections: {{ proxysql_max_connections }}
    interfaces: "{{ kolla_internal_vip_address | put_address_in_context('url') }}:{{ database_port }}"
-   connect_retries_delay: "{{ mariadb_connect_retries_delay }}"
-   connect_retries_on_failure: "{{ mariadb_connect_retries_on_failure }}"
-   shun_on_failures: "{{ mariadb_shun_on_failures }}"
-   monitor_username: "{{ mariadb_monitor_user }}"
-   monitor_password: "{{ mariadb_monitor_password }}"
-   monitor_connect_interval: "{{ mariadb_monitor_connect_interval }}"
+{% if mariadb_connect_retries_delay | length > 0 %}
+   connect_retries_delay: {{ mariadb_connect_retries_delay }}
+{% endif %}
+{% if mariadb_connect_retries_on_failure | length > 0 %}
+   connect_retries_on_failure: {{ mariadb_connect_retries_on_failure }}
+{% endif %}
+{% if mariadb_connect_timeout_client | length > 0 %}
+   connect_timeout_client: {{ mariadb_connect_timeout_client }}
+{% endif %}
+{% if mariadb_connect_timeout_server | length > 0 %}
+   connect_timeout_server: {{ mariadb_connect_timeout_server }}
+{% endif %}
+{% if mariadb_connect_timeout_server_max | length > 0 %}
+   connect_timeout_server_max: {{ mariadb_connect_timeout_server_max }}
+{% endif %}
+{% if mariadb_monitor_connect_interval | length > 0 %}
+   monitor_connect_interval: {{ mariadb_monitor_connect_interval }}
+{% endif %}
+{% if mariadb_monitor_connect_timeout | length > 0 %}
+   monitor_connect_timeout: {{ mariadb_monitor_connect_timeout }}
+{% endif %}
+{% if mariadb_monitor_galera_healthcheck_interval | length > 0 %}
    monitor_galera_healthcheck_interval: "{{ mariadb_monitor_galera_healthcheck_interval }}"
+{% endif %}
+{% if mariadb_monitor_galera_healthcheck_timeout | length > 0 %}
    monitor_galera_healthcheck_timeout: "{{ mariadb_monitor_galera_healthcheck_timeout }}"
+{% endif %}
+{% if mariadb_monitor_galera_healthcheck_max_timeout_count | length > 0 %}
    monitor_galera_healthcheck_max_timeout_count: "{{ mariadb_monitor_galera_healthcheck_max_timeout_count }}"
+{% endif %}
+{% if mariadb_monitor_ping_interval | length > 0 %}
    monitor_ping_interval: "{{ mariadb_monitor_ping_interval }}"
+{% endif %}
+{% if mariadb_monitor_ping_timeout | length > 0 %}
    monitor_ping_timeout: "{{ mariadb_monitor_ping_timeout }}"
+{% endif %}
+{% if mariadb_monitor_ping_max_failures | length > 0 %}
    monitor_ping_max_failures: "{{ mariadb_monitor_ping_max_failures }}"
+{% endif %}
 {% if mariadb_monitor_read_only_interval | length > 0 %}
    monitor_read_only_interval: {{ mariadb_monitor_read_only_interval }}
 {% endif %}
-   monitor_connect_timeout: 6000
-   connect_timeout_client: 100000
-   connect_timeout_server: 30000
-   connect_timeout_server_max: 100000
+   monitor_username: "{{ mariadb_monitor_user }}"
+   monitor_password: "{{ mariadb_monitor_password }}"
+{% if mariadb_shun_on_failures | length > 0 %}
+   shun_on_failures: {{ mariadb_shun_on_failures }}
+{% endif %}
 {% if database_enable_tls_backend | bool %}
    ssl_p2s_ca: "/etc/proxysql/certs/root.crt"
    ssl_p2s_cert: "/etc/proxysql/certs/mariadb-cert.pem"

releasenotes/notes/proxysql-revert-to-upstream-defaults-235abb7fb10bfcfd.yaml

@@ -0,0 +1,53 @@
+---
+upgrade:
+  - |
+    The following ``proxysql`` variables have been reverted to upstream
+    defaults (unless overridden using Ansible variables).
+    In case of ``shun_on_failures``, ``connect_retries_delay`` and
+    ``connect_retries_on_failure`` we still have own defaults for all-in-one
+    deployments - but rely on upstream on multinode ones.
+
+    .. list-table::
+
+      * - Variable name
+        - Kolla-Ansible default
+        - ProxySQL (upstream) default
+      * - connect_retries_delay
+        - 1000 (aio) / 1
+        - 1
+      * - connect_retries_on_failure
+        - 20 (aio) / 10
+        - 10
+      * - connect_timeout_client
+        - 100000
+        - 10000
+      * - connect_timeout_server
+        - 30000
+        - 1000
+      * - connect_timeout_server_max
+        - 100000
+        - 10000
+      * - monitor_connect_timeout
+        - 6000
+        - 1000
+      * - monitor_galera_healthcheck_interval
+        - 4000
+        - 5000
+      * - monitor_galera_healthcheck_max_timeout_count
+        - 2
+        - 3
+      * - monitor_galera_healthcheck_timeout
+        - 1000
+        - 800
+      * - monitor_ping_interval
+        - 3000
+        - 8000
+      * - monitor_ping_timeout
+        - 2000
+        - 1000
+      * - monitor_ping_max_failures
+        - 2
+        - 3
+      * - shun_on_failures
+        - 10 (aio) / 5
+        - 5

tests/templates/globals-default.j2

@@ -45,6 +45,8 @@ mariadb_wsrep_extra_provider_options:
   - "evs.inactive_timeout=PT30S"
   - "evs.keepalive_period=PT3S"
 
+mariadb_monitor_connect_timeout: "6000"
+
 nova_compute_virt_type: "{{ virt_type }}"
 
 enable_openstack_core: "{{ openstack_core_enabled }}"