When Corosync notifies that particular node in its cluster
is dead, rabbit-fence daemon fences the failed node in
RabbitMQ cluster as well:
* It casts disconnect failed_node & forget_cluster_node for
the rest of the nodes in the RabbitMQ cluster.
* Does not fence alive nodes with mnesia running.
* Does not fence already forgotten nodes, that means that
only the first node detected a 'dead event' will issue the
fencing action, while the rest of the cluster nodes will
ignore it.
* Requires corosync compiled with --enable-dbus option,
ensures corosync-notifyd and dbus (messagebus) are running.
* Contains temp hacks in the corosync-notifyd init.d script to
w/a upstream bugs https://bugs.launchpad.net/ubuntu/+source/corosync/+bug/1437368, https://bugs.launchpad.net/ubuntu/+source/corosync/+bug/1437359
* Installs init.d and upstart scripts for rabbit-fence daemon and
enables it after the puppet Rabbitmq class evaluated
Note: system events may be monitored with dbus-monitor --system
Note: If corosync package got updated with apt-get, the corosync-notifyd
service would be affected by the mentioned Ubuntu upstream bugs again
and wouldn't start as a result. Make sure to backup the init script for
corosync-notifyd prior to issue the update and restore it once the
update is done.
Doc-Impact: ops guide
Closes-bug: #1437348
Related blueprint rabbitmq-pacemaker-multimaster-clone
Change-Id: I691363386efe01421acc317ef6371ce45a0d4d11
Reviewed: https:/ /review. openstack. org/108792 /git.openstack. org/cgit/ stackforge/ fuel-library/ commit/ ?id=ad9097a1bb5 5cf035f6afa1d12 afd13bd965f2b5
Committed: https:/
Submitter: Jenkins
Branch: master
commit ad9097a1bb55cf0 35f6afa1d12afd1 3bd965f2b5
Author: Vladimir Kuklin <email address hidden>
Date: Tue Jul 22 22:20:27 2014 +0400
RabbitMQ node resource level fencing
When Corosync notifies that particular node in its cluster
is dead, rabbit-fence daemon fences the failed node in
RabbitMQ cluster as well:
* It casts disconnect failed_node & forget_cluster_node for /bugs.launchpad .net/ubuntu/ +source/ corosync/ +bug/1437368, /bugs.launchpad .net/ubuntu/ +source/ corosync/ +bug/1437359
the rest of the nodes in the RabbitMQ cluster.
* Does not fence alive nodes with mnesia running.
* Does not fence already forgotten nodes, that means that
only the first node detected a 'dead event' will issue the
fencing action, while the rest of the cluster nodes will
ignore it.
* Requires corosync compiled with --enable-dbus option,
ensures corosync-notifyd and dbus (messagebus) are running.
* Contains temp hacks in the corosync-notifyd init.d script to
w/a upstream bugs
https:/
https:/
* Installs init.d and upstart scripts for rabbit-fence daemon and
enables it after the puppet Rabbitmq class evaluated
Note: system events may be monitored with dbus-monitor --system notifyd prior to issue the update and restore it once the
Note: If corosync package got updated with apt-get, the corosync-notifyd
service would be affected by the mentioned Ubuntu upstream bugs again
and wouldn't start as a result. Make sure to backup the init script for
corosync-
update is done.
Doc-Impact: ops guide pacemaker- multimaster- clone 421acc317ef6371 ce45a0d4d11
Closes-bug: #1437348
Related blueprint rabbitmq-
Change-Id: I691363386efe01