Need log_slave_update enabled

[Paul Halmos]

Each of the galera nodes needs to have "log_slave_update" enabled. Currently, the node that gets the write request from the LB will log to it's own binary log. The binary logs can be spread across all 3 nodes making recovery difficult. By enabling this feature, all nodes will have the same binary logs.

[Paul Halmos (paul-halmos)]

More details can be found here: https://dev.mysql.com/doc/refman/5.5/en/replication-options-slave.html

tl;dr
"Normally, a slave does not log to its own binary log any updates that are received from a master server. This option tells the slave to log the updates performed by its SQL thread to its own binary log."

Add:

[mysqld]
log_slave_update=1

[Christopher H. Laco (claco)]

Did you test it with "1", or "ON".

According to the docs and the fun history of these things, 5.5+ should be "ON", and 5.1 was "1".

[Christopher H. Laco (claco)]

Seems like most replication docs and examples still use 1, even in 5.5. And it's plural: log_slave_updates

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/213467

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/213467
Committed: https://git.openstack.org/cgit/stackforge/os-ansible-deployment/commit/?id=41332ab35f5ff667cd5d3fdbecf03ffac0e87147
Submitter: Jenkins
Branch: master

commit 41332ab35f5ff667cd5d3fdbecf03ffac0e87147
Author: Christopher H. Laco <email address hidden>
Date: Sat Aug 15 15:50:10 2015 -0500

    Enable log_slave_updates

    Without log_slave_updates enabled, the bin logs are only written on the
    cluster server that handles the request. This makes database recovery
    more difficult as the bin logs are spread across all servers and must be
    combined manually in order to recover the database.

    Enabling log_slave_updates ensures all nodes in the cluster contain the
    complete bin logs and recovery can be performed from any cluster member.

    Change-Id: I978d6aea93369382faeace96b7865b9e56a7995a
    Closes-Bug: #1479473

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/214576
Committed: https://git.openstack.org/cgit/stackforge/os-ansible-deployment/commit/?id=a3f1b83ab850be5baf6ff3f1e599602fdeda29af
Submitter: Jenkins
Branch: kilo

commit a3f1b83ab850be5baf6ff3f1e599602fdeda29af
Author: Christopher H. Laco <email address hidden>
Date: Sat Aug 15 15:50:10 2015 -0500

    Enable log_slave_updates

    Without log_slave_updates enabled, the bin logs are only written on the
    cluster server that handles the request. This makes database recovery
    more difficult as the bin logs are spread across all servers and must be
    combined manually in order to recover the database.

    Enabling log_slave_updates ensures all nodes in the cluster contain the
    complete bin logs and recovery can be performed from any cluster member.

    Change-Id: I978d6aea93369382faeace96b7865b9e56a7995a
    Closes-Bug: #1479473
    (cherry picked from commit 41332ab35f5ff667cd5d3fdbecf03ffac0e87147)

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: juno
Review: https://review.openstack.org/222198

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/222198
Committed: https://git.openstack.org/cgit/stackforge/os-ansible-deployment/commit/?id=09870e589eaea11e3d5f5d184266d017c3965131
Submitter: Jenkins
Branch: juno

commit 09870e589eaea11e3d5f5d184266d017c3965131
Author: Christopher H. Laco <email address hidden>
Date: Sat Aug 15 15:50:10 2015 -0500

    Enable log_slave_updates

    Without log_slave_updates enabled, the bin logs are only written on the
    cluster server that handles the request. This makes database recovery
    more difficult as the bin logs are spread across all servers and must be
    combined manually in order to recover the database.

    Enabling log_slave_updates ensures all nodes in the cluster contain the
    complete bin logs and recovery can be performed from any cluster member.

    Conflicts:
      rpc_deployment/roles/galera_config/templates/my.cnf

    Change-Id: I978d6aea93369382faeace96b7865b9e56a7995a
    Closes-Bug: #1479473
    (cherry picked from commit 41332ab35f5ff667cd5d3fdbecf03ffac0e87147)

[Davanum Srinivas (DIMS) (dims-v)]

This issue was fixed in the openstack/openstack-ansible 11.2.11 release.

[Doug Hellmann (doug-hellmann)]

This issue was fixed in the openstack/openstack-ansible 11.2.12 release.

[Davanum Srinivas (DIMS) (dims-v)]

This issue was fixed in the openstack/openstack-ansible 11.2.14 release.